KOffice Team: A Handful of Coders, a Lot of Code

nickbrown writes: "In this interview with the KOffice development team it is revealed that only about 4-6 people are working on the suite of applications. It would appear they lack the resources to keep up with the likes of openoffice. Worth a read as it highlights the troubles they are having trying to produce a truly productive office suite for KDE."

Comments

Re:StarOffice?

[fiziko]

The free version of StarOffice is OpenOffice, which was mentioned.

hmm

[perdida]

I am not sure whether the author here is trying to start an office-suite software choice argument or is trying to get people to code for this project.

Exposing the core business weakness of this software development is not going to help it get more work done. It's like putting out a flea-ridden sad puppy and saying, "look how sick he is, don't you want to adopt him?"

Re:hmm

[roguerez]

4-6 people isn't that few. You would like to have people to participate in UI design, icons, etc. But the 'engine' of these applications could be done by a handful of people. Perhaps they want to much and they should concentrate on the word processor. Just an idea.

Re:hmm

[praedor]

The problem with opensource/free software development is precisely because it is so volunteer-driven that work is rarely completed or focused on a polished product. Everyone wants to do the "cool" coding, or "scratch an itch" which all too often merely means: code something that works but is NOT user friendly, attractive, or otherwise appealing to the a wider audience.

Get OVER wanting to code ONLY the cool stuff and grow up, realize that sometimes tedious coding is required (a MUST) to put polish on a product. Get over coding in a "cool" wizbang feature and make it useable and friendly even when it means less fun.

Do the fun coding, sure, but accept reality - that not all the coding that NEEDS to be done is cool but it still must be done. Do it. That is part of being an adult, accepting that everything that needs doing isn't all a party, nor should it all be a big party.

Commercial software developers don't have this problem. The boss says "code this" and you do it and it gets done and you end up with a good product that people can use and WANT to use.

Unreadable

[fiziko]

Does anyone out there want to mirror that in a readable font?

Re:Unreadable

[fiziko]

I use Konqueror, actually. I haven't upgraded in a while, though. (I'd rather not disturb anything until after the final draft of my thesis is finished and submitted.)

4 to 6 employees

[yintercept]

While I wait for the poor slashdotted server to retreive the article, I should mention that the right 4 to 6 people can outperform the wrong 400 person programming department.

Small teams often can focus on the issues at hand and make a more tightly tuned product than the big teams, and if they fail...not as much is lost. This is especially true when you have a good well established foundation onwhich to build.

Re:4 to 6 employees

[Geek Boy]

I don't think you understand the magnitude of this project. It is huge, and very complicated. It takes a very long time just to reverse engineer other file formats and build filters.

There is no way that 4-6 people working in their spare time can complete a project like this. Read the source code and you'll begin to get a real feel for the magnitude of things.

Re:4 to 6 employees

[Otter]

... I should mention that the right 4 to 6 people can outperform the wrong 400 person programming department.

Well, yes and no, depending on how focused the task is. David Faure and some volunteers on the Konqueror team kicked Eazel ($12 million in VC money, 100+ employees) around the block, and the Konqueror + khtml guys did the same to Mozilla. (Yes, I know Mozilla has gotten a lot better, but they still did. Yes, I know Mozilla isn't a browser but a portable development platform -- that's why the Konqueror guys kicked them around the block by making a browser.)

But building an office suite involves a lot of modularized tasks, and there's just no way for a handful of even the best devs to do them all the way a large team can.

Re:4 to 6 employees

[Darren Winsper]

Err...speaking about the web browser part of Konqueror, it has not kicked Mozilla's arse in any way, shape or form. Here's why:

1) Stability. You can not honestly say Konqueror is any more stable than Mozilla
2) Cross-platform support. Konqueror is not available on Windows or Mac OS
3) Standards support. Mozilla has better support for more standards than Konqueror
4) Scope. Konqueror is just a web browser/file manager. Mozilla has a mail client and an HTML composer. Some people may not want them, but others do. I think the mail part of Mozilla kicks KMail all the way to next Tuesday.
5) Tabs :)

Re:4 to 6 employees

[Darren Winsper]

I use KDE, but I'd be using GNOME2 if I could get the problems with the status docklet sorted.

As for Konqueror, I think Mozilla's better.

As for KWin, there's not much that's lighter than Sawfish.

Re:4 to 6 employees

[yintercept]

There is no way that 4-6 people working in their spare time can complete a project like this.

The article says:

David: People usually assume that writing an office suite is very difficult That's not true. _Some_ part of it are difficult (can I say WYSIWYG ? ;), but there are many many missing features that are easy to add.

Thomas: It is not too hard to get in. It is written in a Object Oriented language after all. Most stuff is cleanly isolated from the core and well documented. The hard part is getting to know the structures of the applications, but here, we can help. After that, its just programming, which is fun ;)

When you get down to it, an office product is really just a collection of different objects that have defined rules for interacting with eachother. You don't have to read all 350,000 lines of code to find a starting point for a new module.

It takes a very long time just to reverse engineer other file formats and build filters.

The article confirms your point on the filters. They mention the filters as the one area where they need the most help. It would be nice to live in the ideal world where the producers of file formats create cleaner documentation, or provide industry standard filters for their formats that you can integrate in a product. Unfortunately, the temptation is to use poorly documented, proprietary formats to create a monopoly position in the market.

I didn't say small groups can handle all projects, just that they can do some things very well. I would agree with the article that filters are a rough edge.

Re:4 to 6 employees

[GlowStars]

Save a webpage on your hard drive for future reference, load it in Konqueror, click a link on it and presto, your LOCAL file-path (file:/home/eyal/projectomega/ref.html) is sent as HTTP-REFERER to the webserver. No other browser does that. This tells a lot about how much KDE developers care about your privacy.

Did you sent in a bug report, or even better, a patch? Complaining is one thing, but helping to fix the issues at hand is the better alternative.

Re:4 to 6 employees

[Darren Winsper]

I'm not passing personal experience as facts. There are no valid tests that show Konqueror to have anywhere near the standards support of Mozilla. Where is Konqueror's support for MathML? How about decent DOM1? Has anyone written support for XSLT for Konqueror yet?

I tell you what, go to http://www.w3.org/Style/CSS/ in KDE/Konqueror 2.2.2 and Mozilla 0.9.9. Which renders the site correctly?

Speed? Mozilla's not the fastest, although if you use Galeon it is comparable to Konqueror.

Mozilla 0.9.9 does support font AA.

As for stability, do you have any evidence the MTBF of Konqueror is any better than Mozilla's?

Re:4 to 6 employees

[Darren Winsper]

Konqueror from KDE 2.2.2 is quite stable on my machine. I don't use Mozilla in Linux but on Windows it seems quite stable as well. I don't think one has a huge advantage over the other in this area.
Which was my point, Konqueror does not beat Mozilla in this area any more. Back when I first started using Konqueror, this was true.

And Mozilla isn't available on AtheOS ;-).
And neither is Konqueror, only KHTML.

Your points about Qt are somewhat valid, but you'd need to port KDE to be able to port Konqueror to another platform. Mozilla doesn't need this.

I don't know where you might get data to show this one way or the other
Try here and take a look here. Notice the mouse hover over the side elements doesn't work correctly in Konqueror. Also, how do I select alternate stylesheets in Konqueror?

It works on many sites that Mozilla doesn't even try to run
Such as?

"Just" a web browser/file manager? It rips CDs, it interfaces with digital cameras, it browses Windows networks, it browses the web, it manages files, it does FTP, it burns CDs, it manages MP3s on your Nomad Jukebox, it browses your RPM database, it is totally integrated with KDE and previews tons of file types. "Just" a browser/file manager indeed! I think KMail is a great mail client, and I prefer the concept of Quanta+ to a WYSIWYG html editor.
Fair enough, but at that point you're getting somewhat less impressive, since you're saying "KDE does more than Mozilla" rather than "Konqueror as a web browser spanks Mozilla", which still hasn't been proven.

For the life of me I can't understand what the big fuss is about tabs, but they'll be in KDE 3.1.
I will not use a browser without tabs as long as I don't have to. And as for "It'll be in 3.1", 3.0 isn't even out yet.

Re:4 to 6 employees

[elflord]

I fucking hate the hypocrisy around here. The mono and dotgnu projects are competeing with .NET with small numbers of people generating huge amounts of uber-technical high quality code. Result: endless craploads of KDE fuckwads slagging them off.

I disagree. I'm a Qt user, and I also use KDE as my desktop. I'm also a big fan of the Mono project, and IMO, it's a far-sighted and bold move on Miguels part. I don't think the people bashing Mono are the KDE fans. Most of the Mono bashers are ill-informed zealots who mistakenly believe that Mono is "Microsoft", therefore "bad".

As for Koffice, I wouldn't say they're doing badly in terms of results. Some of the components are doing very well, some aren't. Developing an office suite is an enormous undertaking, and so it's taking time for both the GNOME and KDE projects. As for OpenOffice, it's based on a mature codebase that's been around for years. StarOffice was a usable product before Qt had a public release.

Re:4 to 6 employees

[elflord]

The problem is that for an application as complex as an office suite, you need a whole bunch of people to do things like documentation, QA, build management, project management, etc.

A lot of these extra jobs are created by the overhead of managing several developers. For example, you don't need a special "project manager" if you only have a few developers, but if you have a large number of developers, you obviously do. The same is true for build management. Documentation and QA are important, but they're of limited utility until the code is mature enough that the blatantly obvious bugs have all been fixed.

Re:4 to 6 employees

[NaturePhotog]

The original GEOS development team was probably a dozen people (I'd have to look at some old files to get the exact count). Geoworks Ensemble not only included the GEOS operating system (pre-emptive multi-tasking, object-oriented UI, single image model, etc.), but a word processor, vector drawing program, PIMs, file manager, and other goodies. At the time, it was well beyond the Microsoft offering of Windows 3.0 + MS Works.

GEOS didn't get a spreadsheet and flatfile database until V2.0, and Internet apps until later. I recall one of the 3.5 engineers on the spreadsheet project (1 = FP math, 1 = data storage and calculation, 1 = app framework and UI (me), 0.5 = charting) finding an article about the Lotus 1-2-3 development team, and their engineering team was larger than all of Geoworks at the time.

Sadly, the same miracle of small numbers wasn't pulled off by Geoworks' sales department. Not entirely their fault, as the 'chicken and egg' problem of lack of available apps, and a Unix-based SDK kept most developers (other than AOL née Quantum Computer Services) away, which didn't add many available apps.

The target that KOffice, et al, are facing now is a lot further along (especially Excel; Word still sucks), but the general idea is still true: adding bodies doesn't always make the results come any faster. In the words of Wernher von Braun:

Crash programs fail because they are based on the theory that, with nine women pregnant, you can get a baby a month.

Which isn't to say more programmers wouldn't help, just that they need to be the right programmers.

Re:4 to 6 employees

[Spy Hunter]

And neither is Konqueror, only KHTML.

Which is the web browsing portion of Konqueror (minus bookmark management and various other small UI thingies), which is what we're concerned about here as you point out later in your post. The port to AtheOS proves the portability of KHTML. And Konqueror/Embedded does NOT need KDE, that's the whole point of Konq/E. You wouldn't have to port all of KDE.

Fair enough, but at that point you're getting somewhat less impressive, since you're saying "KDE does more than Mozilla" rather than "Konqueror as a web browser spanks Mozilla", which still hasn't been proven.

Right. These things are all tangential to making a good web browser, as is an HTML editor and mail client and ICQ client and whatever else Mozilla has. I'm not necessarily arguing that Konqueror spanks Mozilla here, either, I'm just pointing out some things.

And as for "It'll be in 3.1", 3.0 isn't even out yet.

Tabs were planned for 3.0 but were too late for the freezes and such, so they got pushed back to 3.1. You can see it on the 3.1 planned features page. Note that David Faure has his name attatched to it - since he's a paid developer he'll surely get it done.

Re:4 to 6 employees

[rseuhs]

And another thing...

The main reason why I prefer Konqueror is because it has far superior bookmark-handling (creating bookmark-directories is much much less painful), and more importantly, it respawns all windows when I relogin.

So when I read slashdot or other forums, I just leave the windows open and they will be loaded just like they were the next day. - No more need for temporary bookmarks.

Re:4 to 6 employees

[Darren Winsper]

Freetype isn't dependent on X. By using FreeType they can port it to other platforms more easily.

Re:4 to 6 employees

[Darren Winsper]

"Does settings->configure konqueror->stylesheet looks difficult ? I would not think so."

That's got nothing to do with what I'm on about. I'm talking about site which have more than one stylesheet suited for different people. How do you select between them in Konqueror?

Re:4 to 6 employees

[Spy Hunter]

Thanks, glad you like it! I guess you get the lollipop.

It was an extra credit assignment for my Discrete math class. We were learning about Pascal's triangle and ways to determine if a number in the triangle would be even or odd. The assignment was to make a program that would print out the pattern of even and odd numbers in the triangle using what we had learned.

I have an even shorter version of the program (106 characters) in C++, but due to the funny way Slashdot calculates sig lengths it won't fit in the sig box. (all the < and > signs must be turned into &lt; and &gt; because of HTML, which count as 4 characters each).

Scratch an itch?

[Anonymous Coward]

Where's the "scratch an itch" factor in coding an office suite? People that are smart enough to contribute to these projects tend to prefer logical markup such as LaTeX. So you'd expect very few OSS programmers choosing to work on this as their hobby.

Re:Scratch an itch?

[arcade]

It seems that you've got an itch to scratch ...

I'm sure there are others that want to scratch the same itch.

a better article would be an investigation of ...

[Anonymous Coward]

...open source tactics.

I mean, like, really.

Why don't open source coders collaborate more? Why in the hell do we need a half dozen different office suites?? Give me one good implementation, and that's all I'll ever use. Is it just that getting geeks to cooperate is like trying to get a group of 3 yr olds (or cats) to all go in the same direction -- near to impossible??

Re:How about collaboration?

[John_Booty]

The project is so big, to compete with MSO, that perhaps all the different groups (OpenOffice, KDEOffice, Gnome office, Gnumeric, Abiword, etc) should all collaborate somehow and make a killer product. Then maybe we'd have a MSO killer.

Sure! Because combining 4-5 totally different codebases that do the same thing always results in a product that's 4-5 times better!

Open Source Leeches

[Anonymous Coward]

This is the real way open source works. Everyone cries and screams about open source, but when the rubber hits the road only a couple of people do a decent job and do the actual work, while the rest of the people just rave on about how l33t an open source coder they are. Most programmers expound open source... then try and figure out ways to sneak the code into their own stuff to make money.

Article

[smak]

This interview was originally planned to take place during the FOSDEM, but we didn't have the time to do it, so it was held on IRC on Tuesday, February 19th 2002. Of course, the log had to be edited and reshaped. The result is this informal interview.

The participants were:

Philippe: Philippe Fremy, (freehackers team), interviewer.

Thomas Capricelli: support, hosting, photos and help (freehackers team).

Ben: Ben Adler, logging and help

Werner: Werner Trobin, KOffice developer

David: David Faure, KOffice developer

Laurent: Laurent Montel, KOffice developer

Thomas: Thomas Zander, KOffice developer

The KDE room at fosdem Philippe: How was KOffice started ?
David: Torben Weis originally started KOffice. He wrote the libs and Reggie (Reginald Stadlbauer) wrote KWord and KPresenter. Torben also wrote KSpread.
Werner: Torben is a programming-ueber-god. He is a programming-task-force in himself. :-) Kalle (Matthias Kalle Dallheimer) added KChart, which was developed for his company (Klaralvdalens Datakonsult)

Philippe: Where are they now ?
David: Reggie is working for Trolltech. He is behind QtDesigner. This is why it had a "Tools" menu for inserting things, instead of "Insert". Tools is what KWord and KPresenter had.
Werner: Torben is working at the Technical University of Berlin IIRC. Last time I talked to him he was developing a CASE tool or so. Nothing KOffice related.

Philippe: How and when did you get involved in KOffice ?
Thomas: Ohh, I don't recall exactly when. Something like 2 or 3 years ago. I picked up where Reggie left off since he ran out of time. KWord was unmaintained at the time.
David: One year ago. KWord was really needing some help, so I jumped on the boat. I proved that it was actually possible to use the richtext stuff in KWord, which Reggie and Thomas had started to doubt, I think. Backporting QRichtext to Qt2 was easy. The real challenge was redesigning KWord around it. Reggie and Thomas had great plans, but no time for them, so I just did it.

David: In general I realized that I never *ever* started anything new in KDE ;) I always joined in the effort started by someone else. I'm the "manpower that makes things possible", maybe, but I never take the initiative to break things myself. I joined Waldo for ksycoca, Coolo for kio-make-it-cool, etc.
Thomas: I like working with David. I have ideas and a general design (plans sounds good) in my head but no time to do it. David is just someone very good at the implementation and the details and stuff; works great together IMO :)
David: I agree :)

Philippe: Could you give us a short report of who is doing what on which application in KOffice right now ?
Werner: I am mainly hacking the core libraries and filter related stuff.
David: I am implementing WYSIWYG in libkotext (common to KWord and KPresenter), and some other KWord fixes. I really thank Mandrake for letting me work fulltime on KDE.
David: Nobody is hacking KSpread right now except some random contributors
Laurent: I improve KPresenter (zoom + page), add the DCOP interface to KWord/KPresenter. I also fix some bugs in KChart.
Thomas: I am hacking the tables in KWord. For 2 versions they have been rewritten twice, lets do it another time. This time I actually think I know what I am doing. So I am implementing it the way I think it should be done. Plus fixing the bugs that bug me in other parts of KWord. Plus I am always bugging David about stuff he is doing in KWord.
Laurent: Nobody is working on kugar and kivio.
David: Kontour is being redesigned by 2 or 3 people, but although they had the time to break everything, it doesn't look like they have time for fixing things now. :) Krayon had nobody for quite some time. Recently someone committed a few fixes, though.
All in all, it's almost one main developer per application only, and random contributors... surely not enough.

Philippe: What about KFormula and KChart ?
David: KFormula has a single developer, Ulrich Küttler with little time. KChart has two parts. The core engine, developed by Klaralvdalens Datakonsult (Kalle's company), and the KOffice part, on which Laurent works sometimes.

Philippe: What about the remaining ones, in the KOffice cvs: KPlato, graphite, kosoap ?
David: those 3 are unreleased alpha stuff.
Werner: graphite is my little pet project, nothing worth writing about in the current shape.
David: koshell is an old hack of mine, I'm trying to pass its maintainance to Sven Lueppken, but it's really Werner who's hacking it for now ;)
Werner: I'm really lacking time for hacking
Thomas: KPlato is a planning application, its been through the design phase. The design is implemented and it is basically waiting for people to finalize the design, I.e. fill the classes allready there. KPlato is 'designed' by 5 people talking on a mailing list without a line of code. I have implemented that design completely. The design I am talking about is just the data classes, nothing more. Naturally the next thing that has to be build after that is a GUI. But that is simply not done.

Philippe: a good project planning application is really something missing.
Thomas: I know, that is why I jumped in, hoping to kickstart the project. But after almost no code was written by others I found that this was not worth my time for now.

Thomas Zander and Laurent Montel
Philippe: So basically, you are 2 full time, paid developers, plus some filter developers, plus 3 or 4 part time contributors.
David: hmm, more than 3 or 4 I think, if you allow "part time" to be very small.
Werner: Yes, the filter developers are Nicolas Goutte, Shaheed Haque (lacking time), Ariya Hidayat, Clarence Dang,...
David: lots of filter developers, if you count them. Usually one per filter.

Philippe: It looks like some help would be welcome.
David: Yes, definitely. On the applications (e.g. Krayon, KPlato, Kontour etc.) we definitely need help. Those apps have almost nobody working on them. The other apps do need contributors, too.

Philippe: What is the programming experience required to contribute ?
David: People usually assume that writing an office suite is very difficult That's not true. _Some_ part of it are difficult (can I say WYSIWYG ? ;), but there are many many missing features that are easy to add.
Thomas: It is not too hard to get in. It is written in a Object Oriented language after all. Most stuff is cleanly isolated from the core and well documented. The hard part is getting to know the structures of the applications, but here, we can help. After that, its just programming, which is fun ;)
Laurent: Any help is welcome, whatever the experience. Some people can create templates or write documentation, tutorials, etc... It's not necessary to know C++. Of course, you must know C++ if you want to hack on KWord.
Werner: There's also a need for GUI guys and artists. We don't have a KoShell icon, for example, and the user interfaces are a bit crowded.

Philippe: Do you have places where you would like to see someone contributing right now ?
David: Definitely. Right now, KPresenter could have a UI redesign (hiding the least used toolbar buttons etc.). KWord needs dialogs for creating envelopes. etc. etc. Many small jobs. A new KWord developer actually started by fixing a simple error message, and is now porting all of KOffice-1.1.1 to KDE3....

Philippe: What are the planned improvements for KOffice ?
David: Right now we are busy fixing the things that we broke. In KWord, due to the WYSIWYG text layouting. In KPresenter, due to zooming support. This sounds small, but actually, it led to a big design change.
Laurent: I am fixing KChart and adding some features. Footnotes for KWord is also planned.
Werner: My short term goal is to implement the rest of the new filter system. The long term goals are e.g. a better clipboard, better support for embedding, a general clean up of the core libraries, ... I also plan to work on graphite a bit more if the free time allows it.

Philippe: Do you have a list of features that will make it into the upcoming release ?
David: Now about the long term plans... Those really depend on how much man power we can get. We have a list for what's already done (see the KOffice release plan, more stuff will be added to that while being done) However, I can't say what will be in the release that is not already done. That's where time / number of developers is the problem.
Thomas: My long term plans for KWord (which is all I am restricting myself to at the moment) is adding more stuff that makes it easier to use the application. Like markup macros. And character styles, frame styles and page styles..
David: My next step for KWord is frame z-order, and then looking at the buglist to see what people requested (for instance, double underline...)

Philippe: You planned 3 Beta and 1 Release Candidate, isn't that too much ? KDE3 had originally 1 beta and 1 rc (which turn into a second beta), no ?
David: It's exactly what we had for KOffice-1.1 . The release plan for 1.2 is _exactly_ the same as the one for 1.1, just one year later. Code reuse, release plan reuse ;) I think KOffice-1.1.1 is only as stable as it is due to the multiple betas
Werner: KOffice gets very few testers. Some people download it and work with it for five minutes. This simply doesn't highlight nasty bugs, but just real showstoppers. There are only few people actually e.g. writing a longer letter or creating a whole presentation.

Philippe: So you need more users too ?
Laurent: yes !
Werner: Definitely. Most of the testing is done by us developers
David: Actually, betas are a really great thing for developers. We can release code for people to test, without having to worry about it being considered a final version, and hence a must-be-completely-stable one. More developers is much more needed than more users. Not handled bug reports are the proof of that. But it's true that more (advanced) users helps finding the source of some problems. Sometimes.
Thomas: We have a number of end-term students (last year project in the company) basically they all say that KOffice is not usable at the moment, most bugs are being fixed, when that is done the not-enough-users problem will go away.

Laurent Montel and David Faure hacking
KPresenter during a boring presentation
Philippe: The must-ask question: what is the state of the import/export filter for MsOffice ?
Werner: We have some import filters, at least. The WinWord import filter imports tables and text/basic formatting (pictures are disabled). The Excel import filter works quite nicely.
David: There is currently no export filter for MsWord. This is an area that definitely needs help.
Werner: As I worked on that myself I can assure you that exporting to WinWord is *not* funny.

Thomas Capricelli: Could we have some more information about picture embedding in KWord ? What's the problem ?
David: Actually, the problem is for the Windows Meta Files, which map to a Kontour part. Since KIllustrator (now called Kontour) crashed for some time, importing wmfs was disabled. And since Kontour is currently very broken (redesigned), enabling it now wouldn't do much good :}
Thomas: Picture embedding has gone a long way since the horrible support in 1.0. Pictures are correctly scaled, correctly and intuitively placed and can be embedded or placed external of the main document. Lots of stuff has to be done, but it is getting there quite rapidly.
David: I think it's only WMF embedding that is disabled. Importing MSWord an document with bitmaps should work ok even in 1.1.1

Philippe: I remember the guy of wvWare starting a project for creating a generic Word import/export. Did that work ?
Werner: This "guy of wvWare" was me :) ...and as noone except Shaheed joined, it somehow didn't really work out. Some code is there, but I just don't find the time
Philippe: And on the side of abiword ? Weren't they supposed to help out ?
Werner: The abiword guys said they're going to help, but noone really did. They were also lacking time. I've written 99% of all code in wv2 ;)

Philippe: Can you reuse the stuff from OpenOffice ?
Werner: It's really hard. They use libc and nothing more. They have their own toolkit and stuff. You can't just take parts of it and reuse it. And the design is, well, hard to follow.

Philippe: But they somehow managed to understand how MsWord works, so I imagine you can at least get hints from there ?
Werner: yes, it helps, but it's very time consuming.
Philippe: What is the difficulty with writing it ? Is MsWord's format so messy ?
David: Afaik one very messy part is the incremental saving, i.e. appending the changes at the end of the doc instead of modifying the real data.
Werner: They don't save the full file all the time, but append "diffs". For example, I spent 3 days hunting a bug in my filter. The bug was in the specification.
David: I also suppose that writing an export filter for a closed commercial app is awful - if it crashes or says "can't open", you have no way of debugging that The format is just partially documented and somehow broken, not fully secret & broken.
Werner: There are situations where WinWord just crashes without further notice if one bit(!) is set incorrectly ... incorrectly as in "according to the SPEC" ;) It's really not funny to develop that, I can assure you. So when you don't get paid for it you lose the motivation and move on to something more interesting.
Philippe: We sure all can understand that.
David: Sometimes I wonder if I shouldn't spend some time continuing the wv2 stuff.

Philippe: I did not pick it up: you are working on the filter directly in KOffice or working on wv2 which will be used by KOffice ?
Werner: Both, but I am lacking of time. Just to give you some numbers, StarDivision had 2 full-time developers for the im/export filter for more than a year! Now guess why KWord can't export .doc :(
David: AFAIK Shaheed started to work on wv2, then seeing that no abiword developer could help making this advance, went back to his KWord msword filter....

Werner: The problem isn't (at least for the import case) that the file format is ultra secret/broken. It's just lack of time. It's so frustrating to read a .doc file in the hexeditor for hours
Philippe: Yes, not very funky stuff to work on !
Philippe: Some distro should give a kick to that ! I can't think of any Redhat support, but we could contact Suse.
David: Mandrake already sponsors two fulltime KOffice developers. Maybe it's time for the other distros to give a hand ?
Werner: Caldera sponsored my development last summer. I had six weeks to work on wv2. All the code which is there was written in that six weeks :)

Philippe: Everyone seem to be thinking OpenOffice is the Office Suite of choice for Linux. But I am not sure it integrates as nicely as KOffice applications do.
David: OpenOffice is not the Suite of choice, it is rather the only one that's currently doing the job. If KOffice improves, everyone (almost) will prefer KOffice (lighter, more integrated with the rest of KDE, nicer toolkit to use IMHO, etc.)
Thomas: After hearing 2 students swear for 3 days, I'm pretty sure Openoffice will not be the solution for companies for some time to come. (They switched to LaTeX now)
Philippe: KOffice has much more potential: it was developed by very tiny manpower in comparison with other suits or applications.
David: Yes, the power of KOffice is that it fully reuses the power of Qt and KDE. KOffice can embed documents thanks to kparts, it has configurable toolbars thanks to the XML-GUI, etc. This means KOffice development can concentrate on the actual office functionality. The rest is provided by Qt & Kdelibs.
Werner: Some people compare the pure line counts. OpenOffice has approx 8million lines, KOffice about 350.000 or so. We can use a lot of nice stuff from Qt and KDE. OpenOffice has to invent *everything* above libc. For a good example please see David Faure's 8-lines text editor, or the KPart article
David: I wrote a 5 lines "generic viewer". It can show any kind of file for which there's a KParts part. See docu in componentfactory.h ;)

Thomas Capricelli: Did you read about latest RMS statement about KDE/Gnome cooperation ? What do you think about it ?
David: this doesn't apply to KOffice, since there's not really a gnome office suite. :)) But I find this very good that RMS states this. I think this 1) finally gives some credibility to KDE for the "GNU only" people. 2) it might actually foster some cooperation.

Philippe: Is the Euro symbol supported ?
Laurent: The Euro works in KOffice no problem !
David: Yes. see this link on the KOffice website.

Philippe: And i18n support ? Are various exotic languages correctly supported ?
Thomas: Yes, see this example.
Laurent: Yes, now KWord supports bidirectional languages => arabic language etc ...
Werner: Qt (thus KOffice) supports UTF-16 and we use QString for everything the user sees. The translations are done by the KDE translation team (http://i18n.kde.org)

Philippe: I wonder if Abiword does that too. Do they use Pango and Gtk 2.0 ?
Werner: Yes, Abiword does that too. At least Dom Lachowicz, the guy I talked with about wv2, was implementing that. But they do their own Right-to-left support, from what Dom told me. I have no idea if that is in the latest stable release, though.
Laurent: Abiword is not integrated in Gnome. It uses some gnome libraries but not much.

Ben: Doesn't Microsoft want to use xml file formats in future office formats? If yes, wouldn't that make import/export a lot easier ?
David: Good question. It mostly depends on how they do it. The main advantage of XML is that it's readable. At least that will help. As for conforming to any spec of theirs, we know how bad they are at that. They will surely "extend" XML to do things their own way.

Ben: What about Aethera in KOffice? As a user, I'd LOVE to see Aethera in KOffice. How much do you trust theKompany to actually deliver Aethera, would you put it into KOffice CVS if it was an open license ?
David: Aethera is a mailer, right ? Then it has nothing to do with KOffice. It's important to understand what KOffice is about. It's about editing *documents*, and being able to embed documents from other KOffice components. There are plans for more KDE PIM stuff, integrating kmail with contact / calendar stuff. The kdepim module in CVS. This needs help too! KOffice doesn't mean "stuff you do at the office" in general. KMail is in KDE, but not in KOffice.
Laurent: Nobody works on Aethera ! At least publicly.
Ben: The latest aethera beta is 4/2001!
Werner: They also have Kivio there, but it doesn't seem to be maintained either.

Philippe: Where would you like to see more cooperation with other projects ?
Laurent: Filters, I think.
David: Yes, filters definitely.

Philippe: Do you think it would be possible to define a common free format for Office files, like it was done for .desktop files ?
Werner: The first step is the common packaging format. This is discussed on the office_standards list, on their website. But I don't know the current KWord DTD good enough to participate in such a discussion.
David: We are currently discussing with the OpenOffice hackers about the ZIP packaging. So the thing we can say in the interview is that one of the plans for KOffice is to maybe switch to ZIP instead of .tar.gz for two reasons:

for efficiency reasons (ZIP allows "load on demand", "partial saving")
for compatibility with openoffice.
The other plan would be to use openoffice's zip code, and this is what we're discussing with them currently. This solutions makes loading/saving in KOffice much more efficient than using an external ioslave.

Philippe: Are you actively participating ?
David: Maybe we should get more involved in discussing this. You can say this interview will have been useful, I'll discuss the common file format with openoffice in the near future ;)
Thomas (Capricelli): We are glad for that :)

Ben: can KOffice be compiled for embedded devices? Do you think making KOffice available for embedded devices would noticeably increase KOffice's user base ?
David: I think this is technically possible. I don't think any of us will take the time to do that, but anyone with interest in this can surely try.
Werner: It's probably no big deal to get it to compile, but it's hard to strip down the UI that it's usable.

Philippe: Thank you all for participating!

Re:Article

[Thing 1]

Right now, KPresenter could have a UI redesign (hiding the least used toolbar buttons etc.).

Might not be a bad idea to add a "logger" so that all features used are logged, their frequency, etc. This could be used by the developers to fine-tune the displayed features, and hide the lesser-used ones.

Of course, this should be disabled by default like the Mozilla auto-loader, but it could be a prominent option in the installation (again, like the Mozilla auto-loader).

Re:How about collaboration?

[yintercept]

collaboration also has its cost. To make collaboration work you have to devote more time to the logistics of the collaboration than the current team is spending on coding. I love the idea of a team putting out an office suite with 350,000 or less lines of code. The effort has a good chance of capture the fast, efficient niche of the market. PS: I must admit I am envious of the people on the project.

Unrealistic Goals

[lkaos]

The Free Software community is more than capable of delivering an office suite. 4-6 individuals in not a bad team, but they should be focusing on a specific component instead of trying to do it all at once.

Abiword is an excellent word processor and Gnumeric is a great spreadsheet program. Gnome's figured it out. No one wants to work on a large, bloated project for free. Break it up into littl projects and you'll get a couple 4-6 individual teams.

Of course, the article is /.'d already so I am only guessing...

Re:Unrealistic Goals

[Geek Boy]

KOffice is a set of components, not a monolith. KDE uses a component based development model. Thus each of these "applications", kword, kspread, kpresenter, etc, are kparts. You can work on one and never even see the source for the rest. They are just released together as a bundle by the KDE project.

Re:Unrealistic Goals

[Peaker]

Actually, from my experience, programming in a BFS (breadth-first) manner is more productive and motivating than programming in a DFS (depth-first) manner.

You get to see more results and get more satisfaction sooner by showing fewer capabilities in many fields, compared to showing many capabilities in fewer fields.

Re:Unrealistic Goals

[Stary]

Of course this flies straight in the face of accepted wisdom that it's better to have an app do a few things well than to have it do many things poorly.

Oh yes, like Mozilla? Or Office? Or let's even say GCC (which now is GNU Compiler Collection, not GNU C Compiler, because it does so fscking much different stuff noone really knows anymore I'm sure). Just because people repeat it endlessly doesn't make it "wisdom" (though maybe it becomes "accepted"). Wanna talk emacs? vim? Just about the only good app I can think about that kept this motto is Windows Notepad.

Fact is, people using computers today WANT bloated suites of does-everything-apps. M$ Office is popular just because it provides everyone with that little obscure feature that they'd been looking for (but everyone wants a *different* feature).

Now, when it comes to BF or DF programming, alot of programming methodologies (UP, X-treme Programming) recommend programming wide and shallow, and in an iterative way. This actually has nothing to do with the feature set of the app, just of the order in which things are implemented.

Re:Unrealistic Goals

[SurfsUp]

KOffice is a set of components, not a monolith. KDE uses a component based development model. Thus each of these "applications", kword, kspread, kpresenter, etc, are kparts. You can work on one and never even see the source for the rest. They are just released together as a bundle by the KDE project.

That's exactly the problem. Break each application out as a tarball instead of requiring people to deal with the whole bundle and the project will take off.

Mind you it's impressive what's been accomplished with a tiny team, however koffice is in no way usable, I know, I've tried. It's pretty but it doesn't do the job.

Re:Unrealistic Goals

[joto]

Just about the only good app I can think about that kept this motto is Windows Notepad.

Except that notepad doesn't really do anything really well. It doesn't edit large files. It doesn't work well for editing human-readable text as it doesn't have wordwrap. It doesn't work for editing programming languages, as it doesn't handle indented code very well. And it's general lack of features make it almost unusable to anyone having tried a different text-editor.

Re:Unrealistic Goals

[mark_lybarger]

i agree with your point on the usability of koffice. i'm fairly happy with using staroffice 5.2 for any office type work i need to do.
but.. i can't agree with your comment on breaking the source out. i assume you've installed kde itself from source? i have 13 tar files for my kde install which includes kdevelop and koffice. i know there are still a few bandwidth challenged individuals out there (i feel your pain), but i would really prefer a kdecore and kdeother tar that is all encompassing.

Re:Unrealistic Goals

[lkaos]

One cannot compare GCC and Emacs to KOffice and say that it's all bloated software.

GCC and Emacs do so much because they are literally decades old! When they first were developed, they were small, useful tools. Noone sat down and said, "Let's write a compiler to work on fifty million platforms." It just grew to be that way since it was founded on a great architecture (Yes, besides being an asshole, RMS is also a great programmer).

There's a reason XEmacs and GNU Emacs split. Lot's of Emacs folks didn't want a more bloated Emacs.

Mozilla is a special case. The reason it is so bloated is simply because of the nature of what it's trying to do. HTML is so obsfucated that it takes an aweful lot to handle it well. Remember too that Mozilla grew out of Netscape which is a pretty well established code base.

I'm all for developing good architectures, but goals have to be realistic or software just fails.

Are they supper coders or

[Forge]

are they using supper tools.

Remember that these are the same guys who had a hard time with Corba and had to invent DCOP and KPARTS.

I would say that KOffice being anywhere withing sight of the other Office suites is a prety strong testiment to the quality of the KDE Libraries and foundation.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:How about collaboration?

[SerpentMage]

The problem is that the OpenOffice people and many others are copying and not thinking of doing things differently.

For example when I developed my DeNotes I wanted to solve some problems. So I stepped back and said, Ok if I could how would I really do this?

Maybe if they thought like that it would change how we protray office applications. Sure it would be a hard learning curve for the user. BUT it is that or not being able to keep up.

Re:But why?

[PoiBoy]

Applix (still around?)

Applix is still available, but my impression (based on past /. reports) is that the parent company isn't marketing it very heavily or planning on making major new releases. I think the real problem is that they got clobbered by StarOffice and all the other free packages available.

It's a shame, really, since I have always liked it better than all of the other office suites available. Fortunately, the current version is very mature, so the lack of future development isn't a problem -- I'll be using it for years to come just the way it is now.

Re:Concentrate on useful apps for 90% of people

[praedor]

Here we go again. I am once again going to mention something that NO ONE seems to understand is IMPORTANT in a wordprocessor. CITATIONS and REFERENCE HANDLING.

There is NOTHING that really separates one of these wordprocessors (abiword, kword, openoffice/staroffice) from one another. They ALL do the same sorts of things in the same ways...but not a one of them has something a real professional writer REQUIRES. None of them has what ALL college students (and many highschool kids) REQUIRES: an ability to handle citations and build reference pages.

Word and Wordperfect have this on Macs and Windoze via third party apps like EndNote, because the processors were designed so apps like EndNote could seamlessly become part of the wordprocessor (for all practical purposes). Endnote adds its own menubar entry so you can insert citations into your text as you go and then it AUTOMATICALLY generates correctly formatted reference page(s) based on user selection.

There is one and ONLY one "wordprocessor" app in linux/unix that can handle this well and that is Lyx. Lyx is IT. The ONLY show in town for any student or professional writer who needs to cite references in their writings. This is ALL scientists and ALL college students and anyone else who wants/needs to write serious research papers. You MUST cite references or your paper is unfounded heresay, yet NO wordprocessor project is even TRYING to add this capability to their app.

Obviously, none of the projects is interested in any professional writers. Their targets would appear to be letter writers and writers of any work that needs no support (fiction). This is fine, such as it is, but no one gets through this life without having to write SOME papers that require citations and reference pages damnit! If you actually have managed to avoid this then your degree must be astrology or basket weaving.

Someone on ANY of these projects: OpenOffice, Abiword, Koffice...provide either a pipe (like Lyx uses to communicate with the most excellent Pybliographic) so third party apps can work seamlessly with your wordprocessors and insert citation markers and handle reference page generation OR add this capability directly into your wordprocessors! Hell, OpenOffice/StarOffice is PARTWAY there but they dropped the ball BIGTIME. OpenOffice/StarOffice actually has what it calls a bibliography database but it is totally useless. All it will do is store whatever you enter into it MANUALLY. Then, it wont use this database to allow you to insert references/citations into a document, no, it just holds the information uselessly. Go that ONE extra step and make the bibliography database USEFUL: it should be able to import and export all the major citation formats (bibtex, refer, pubmed) and there should a menu item in the wordprocessor that talks directly to this database and will allow the writer to insert citations. At the end, you should be able to hit a "create reference page(s)" button and have staroffice/openoffice add a formatted set of reference pages to the ass-end of the finished document just like Endnote plus Word or Wordperfect will!

I mean, c'mon! Quit focusing on toy writing and add something that hardcore researchers and all college students require. If Lyx can do it, so can the more user-friendly wordprocessors like Kword, Abiword, OpenOffice, etc.

Because of this fatal failing in ALL available linux WYSIWYG wordprocessors, I have to keep Lyx around so I can do my real writing, leaving the rare note or letter to Kword or OpenOffice. What a waste. A nice, BIG app that is kept around just for simplistic writing. Lyx is OK but damnit, I want and need to write, not learn a programming language (latex). My job is to do biological research, not learn programming or hundreds of obscure latex commands just to publish. Lyx and I get by but there needs to be more options for the research paper writer/college student/scientist.

Re:a better article would be an investigation of .

[Fnkmaster]

Software developers rarely agree on things. I've seen fights (usually relatively calm and technical ones, but sometimes escalating to heated yelling and nearly to blows) in engineering team meetings. The only thing that forces people to work together at a company and come to agreement on some idea (often a shitty one, but decided for them nonetheless) is MONEY. Quite simply, your employer is paying you. Sometimes you just have to give up and accept that the motivations of the source of the money need to be accepted if you want them to keep giving you all this nice cash, rather than your own concepts of what is nice, aesthetic or well-engineered.

Open Source projects are motivated by ego in some cases ("I want to bulid the next great window manager") or some sense of technical correctness in others ("I hate the way OpenOffice looks/I hate Gtk widgets/etc."). So there's no real incentive to work together on a bigger project - why would you want to say "yeah, I built some tiny component that's part of this megalithic Open Source project, UltraOffice" when you can say "I am the lead programmer on KWord".

So knock closed source software all you will, money can be an effective way to motivate people to cooperate on bigger software tasks and put differences aside to achieve an overall better result. And though some Open Source business models make this possible, for lots of products (like office suites), nobody yet has figured out how to do this and it very well may be impossible.

Re:Concentrate on useful apps for 90% of people

[Geek Boy]

Here's a great example of the benefit of kword. Launch kword and then run "dcop". You will see kword-xxxxx processes listed. You can then see all the documents in each of those processes.

There are many functions available for communicating with a kword document that is open. Something missing? File a wishlist or contact the mailing list. I'm sure it's easy to add.

Graphically, you can do this with the "kdcop" tool too.

yeah but...

[jefe289]

Koffice is pretty damn good as is... but unfortunately, there are some major insufficiencies: notably, filter support. I think most people who have actually tried using staroffice/openoffice would agree that there is a lot of unnecessary bloat. Another problem, for KDE-zealots like myself, is that other office suites don't integrate well into the slick, advanced, well supported, and popular KDE.

It just seems that with all the different projects out there that are supported by thousands of developers... there are only a few people for each that actually use the program. However, I see K-Office getting a lot more usage. It just seems that you get a LOT more *useful* code by improving k-office than by improving xyz-random project on Freshmeat. (my 2 cents)

Re:How about collaboration?

[Rentar]

Sure! Because combining 4-5 totally different codebases that do the same thing always results in a product that's 4-5 times better!

Wasn't that the way Emacs was made?

Re:Concentrate on useful apps for 90% of people

[zander]

Don't worry praedor, you are still on my TODO list :)

Maybe you missed the fact that KWord does have plugins like you seem to require. The latest version has a thesaurus plugin for example.

Anyway, exactly _this_ is why more authors are needed. The basis is there; just the work needs to be done.

Too hard to get started

[SurfsUp]

"In this interview with the KOffice development team it is revealed that only about 4-6 people are working on the suite of applications.

Yes, well I know exactly why that is. A few years ago I decided to hack kmail a little, until I found I had to build pretty much all of kde from source to do it. As opposed to installing the -devel headers as you would for more developer-friendly applications, then just untarring, ./configure, make. This stopped me, I see no reason why it wouldn't stop lots of folks.

Casual hacking is the way to get started on a project, it's wrong to require a whole huge cvs import and hours of mucking around with scripts trying to get the thing to build. Once you start with the casual hacking and submit a few patches, it's much easier to justify the effort to get in all the way and sync up to cvs.

If the koffice team wants more helpers, then they should put the effort into making it easier to get started. That means writing some scripts to pull tarballs out of cvs and hacking together some autoconf stuff. This effort will for sure pay off. People will start sending patches, and after a while some of them will get involved in a more hardcore way.

Look, why are the 1,000's of people hacking the Linux kernel tree? Because you can just grab the tarball and build it, no fuss no muss. Only super-hardcore developers or paid employees are going to get into a project by syncing to the cvs from the word go.

David, bless him, didn't get this 3 years ago, I hope he'll think about it now.

Re:Too hard to get started

[SurfsUp]

http://www.kde.org/anoncvs.html

There are scripts linked to at the bottom of the page that do everything for you as well (one is in webcvs.kde.org).

I know about that page and those scripts. It still takes hours, if you don't believe me try it. (Make sure you're a newbie to kde building first, otherwise your data is suspect.)

Tarballs generally unpack and install in a few minutes and the compile generally works the first time. I'm sorry, but cvs + futzing with scripts that are broken half the time is just not like that.

Re:Too hard to get started

[SurfsUp]

I never used the scripts, I just followed the instructions. Pretty easy, although it did take a long while to download and build. Never used cvs before then.

http://www.kde.org/kde2-and-kde3.html

http://www.kde.org/install-source.html

Helped also. Persevere and it will pay off. You learn a bit about cvs, about writing scripts and you get latest KDE which is always better than releases ;-) You can always get help on mailing lists or from google if you have trouble.

Sorry, this just isn't as easy or fast as:

1) install the headers
2) grab the tarball
3) unpack the tarball
4) ./configure
5) make

And puts way less load on your disk. Think about it: if it was so easy, then why are there only 4-6 active contributors?

I see a some people making the mistake of thinking that cvs and tarballs are mutually exclusive. They're not. Every so often, preferably at some point release a core developer runs a script to pull tarballs for each app out of the tree. If this isn't possible then the project framework is seriously borked anyway.

Then people who want to contribute have the choice of sending patches against the tarballs or joining the hardcore team and syncing up with cvs. Which isn't as good an idea as it sounds, it's not particularly easy to do productive work when the whole system is changing all the time, and maybe you're not a core developer so you don't know what's happening, why or when. Much easier to work with a stable point release, then when you have something you like send the patch to somebody with cvs access. If they blackhole it, find another project, if they accept it, you're on your way, one day you'll have cvs.

It's a big mistake to make every possible contributor go through the same pain that a core developer has to, to sync up with the bleeding edge.

Re:[Small groups are ] Gibberish

[yintercept]

Unless you can identify documented case studies supporting your implausible thesis

My implausible thesis (that small groups can outperform large programming groups) has a large number of case studies and research. I apologize that I do not keep a detailed list of every thing I've read so I can whip out the case studies.

This subject was the main thesis of Frederick Brook's Mythical Man Month. Adding man months (i.e. more people) to projects does not always increase the speed of development. This phenomena has been reported on by other noted authors such as Edward Yourdan. Economists refer to this as the law of diminishing returns. I recall that both Yourdan and Brooks pointing to case studies. I did come across several studies on this topic in the IBM System Review, but don't have them on hand right now.

I did not say that there is a one size fits all solution to the size of development teams. Simply that small is not a bad way to go. Sometimes adding people increases productivity, some times it just adds overhead.

Again, this is called the law of diminishing returns. I can find case studies for you, but that seems like a trivial waste of time for a slashdot thread.

There have also been case studies which show extremely wide gaps in productivity between programmers. Again, I don't have these on hand, and can only point to my experience that some people get a lot more done than others.

I certainly did not say that a small team is guaranteed of success. Just like big teams are not guaranteed success. Now lets get back to the article. What it says is that a small programming group is working through the code of existing products and adding a few enhancements here and there. They are enhancing some objects, adding a few new objects. This all seems quite doable. I have added enhancements to programs with 100M lines of code that, according to the tests, worked.

Using small focussed teams is a proven and cost effective means of programming. It is one of the reason why people both with that object oriented nonesense that you read about in books. If I have to cite case studies, I probably could waste a few hours and fill the board with several hundred. But that seems like a waste of time.

BTW the article mentioned that they had problems with integration issues, like filters, which generally require a great deal of testing and man hours. Some areas of development require more man hours than others.

Re:How about collaboration?

[alext]

Not quite sure how this apparently deliberate misunderstanding has achieved a score of 5 - to my mind the parent statement is clear: collaboration means collaboration, not necessarily convergence by merging code.

Re:[Small groups are ] Gibberish

[swb]

I wonder how the small v. large debate is influenced by management philosophies and organization.

In the limited experience I've had with software development one of the biggest limitations on software development was that merely adequate developers slowed down the really good developers. Projects couldn't continue forward until certain development milestones were met, so good people just kind of shut down.

This effect was magnified by well-intentioned managers. They wouldn't do anything to try to improve the laggards development skills or pacing because they were otherwise meeting the basic goals, were likeable people, and so on. The lack of vulnerability of the laggards also prevented the better developers from taking over the laggards projects or assisting their speedy completion, since the laggards had a sense of ownership and management "backing" which enabled them to maintain control of their segments in spite of the overall degredation of the project.

It leads me to wonder if development groups have ever applied a 6-Sigma style management process where the laggards were cut and new people brought in. I understand there's some risk -- new people slow everything down getting up to speed, but it has the potential advantage that it eliminates *known* hindrances, and its not an attempt to increase team sizes. In other words, you're trying to fix the team not make it bigger, hopefully avoiding some of the Man-Month style diminishing returns.

Exactly.

[Nailer]

I don't want an office suite for KDE. I want an office suite for Linux.

If there's any advantages exposed in having an application written by a particular toolkit running under a particular environment, then that is the problem. We already have standards for drag and drop, and window manager hints. How about putting 4-5 developers on to the task of making Linux apps act like Linux apps, with the same look, feel, drag and drop, shortcut keys, mime types, etc. Keep the choice, just allow standardized configuration by default.

Re:Exactly.

[scrytch]

How about putting 4-5 developers on to the task of making Linux apps act like Linux apps , with the same look, feel, drag and drop, shortcut keys, mime types, etc. Keep the choice, just allow standardized configuration by default.

I believe there is such a project. It's called KDE. And there's another one called GNOME. Oops. Back to the drawing board.

Frankly, get the keyboard accellerators standard and most of cut/paste interoperating, and you'll be fine. It's not like people expect to be able to drop an excel spreadsheet range into a wordperfect doc after all. People will take an 80% solution if they don't really need the missing 20% (hell, look how well linux has done without acl's standard). Themes don't need to be perfectly consistent so long as the metaphors are consistent. Even microsoft isn't obeying that consistency principle itself (WMP 7 looks like nothing else on win9x)

It's not the consistency (people don't expect it these days), it's not the documentation (people don't read it), it's that the average user isn't being offered a value proposition any better than what they're using now. So you have an office suite and a web browser and chat programs ... which is precisely what you get with windows as well. Not one end user gives a tinker's damn about the source code. If they did, they wouldn't be end users.

It's things like the outrageous pricing of Office XP that might really give the free alternatives a boost. Too early to tell...

Aethera and Kivio updates

[tongue]

The article incorrectly stated that Aethera was not being worked on, and that the last beta was in april of 2001; in fact, there have been at least two betas that I know of, just in 2002. And Kivio is not only being worked on but is supported under windows as well, in the commercial version.

Re:a better article would be an investigation of .

[electroniceric]

I pretty much agree with you, but I'd add a twist.
It's funny, a few months ago, I was saying that open source would never have the polish of commerical OSes like OS X or Windows, because nobody's paid to do the dirty work of making sure that things work consistently across menus, then dialogs, then apps, then the entire package.
My rationale was like yours, people who are in it for free want to do the interesting part, not sweating it out with a hex editor as the article mentions.
Lately, though, I've been muddling on the kde-usability list, and am really quite surprised how much work people are willing to put in to polish up the desktop. Who knows how far the efforts will get, or if people will tire quickly, but at least for now, there seems to be a scratch for every itch. Maybe those open source avatars were right after all.

Re:a better article would be an investigation of .

[Chris Johnson]

There's always a scratch for an itch.

The deal is, with Free software, those scratches can HOOK UP and be built into something huge- eventually.

The tide does not gallop.

Re:Can you say naieve????

[Ogerman]

Can you say, "I can't think for myself?" Get your head out of your ass and look around. Open Source work is freelance by nature. There is no "office development" needed. If you pay some freelance coders to quit their day jobs and work full time on a project they love, they'll turn out some beautiful results for you. Those percentages you randomly quote are total bullshit manager-speak and do not reflect in any way how the Open Source community operates. When you turn out open code that people want / need, it grows automatically much like a seed planted in fertile soil. So that $1 million, divided by say.. 15 programmers for one year, gets the code flowing quickly which in turn creates incentive for other people to make contributions and the momentum builds. After they year is over, the project can stand on it's own. Because office suite software is in far greater demand than web servers, such a project stands to do at least as well as Apache or other highly successful open source projects.

And by the way, it's spelled 'naive.'

Re:Some thoughts

[Ogerman]

The poor Mozilla guys! They're criticized around the block for abandoning the Netscape 4.x base and starting from scratch, and this guy says that they took too long because they had to deal with the Netscape codebase

They did at the beginning. Yes, I'm aware that by now just about everything has been re-written from scratch and that's a good thing. I was just pointing out that during the transition phase and the establishment of the mozilla.org project, there was a lot of time wasted hammering around with old broken speghetti code. That is what the Open Office project seems to be dealing with right now. It's pretty near impossible for someone not intimately involved with the project to be of any help in the development, hence the first goal of the developers has been breaking the bloated monolithic kludge into modules that are actually managable. Sounds like developer hell to me.

Re:How about collaboration?

[John_Booty]

Point taken. :)

I was poking fun at the original post because in my mind, for the disparate groups of developers to band together and make The One Office Suite to Challenge Microsoft Office, they'd either have to start (largely) from scratch, or basically take one of the existing open office suites and all work on that from now on.

It's not that the original post was wrong... as a matter of fact, it was probably right, but it was such an oversimplification as to invite a little fun-poking. Saying "all office suite developers should collaborate to produce a true competitor to MS Office" is like saying "We should all be able grow wings and fly and beer should be free". It's right, but how should it be done?

What a mess...

[Otis_INF]

4-6 developers for such a large project is a very small team. However, when tasks are defined well and goals are set with care, you can do a lot with 4 to 6 people. However, after reading the interview, I got the feeling that KOffice is currently being worked over by some demosceners without any big plan nor well defined tasklist. How can they expect people to jump in when the team itself doesn't sound very solid? (For contrast: in the teams I've worked in in the last 10 years, things were planned, done for a reason, people worked on stuff that they knew was filling in the blanks of the total project. This is IMHO not the case with KOffice. Very sad because I don't think that all this effort these programmers are putting in is paying off in the end.)

Netscape is still a better EMailer/Newsreader

[rseuhs]

4) Scope. Konqueror is just a web browser/file manager. Mozilla has a mail client and an HTML composer. Some people may not want them, but others do. I think the mail part of Mozilla kicks KMail all the way to next Tuesday.

Funny, I think that Netscape 4.7 is still a better newsreader/emailer than Mozilla.

Why you ask?

Simple: Because it doesn't force you to use the very, very moronic layout where you have to have the message occupying the lower half of the window. This is moronic because messages are usually truncated at 80 chars (which is moronic too, but that's another thread) so you have lots of empty space in the message area, yet it is not tall enough.

Netscape4 allowed you to open messages in another window so you could put this window beside the main window to be as tall as the screen.

Yes I know Mozilla can also open it in another window, but that's 100% useless because it doesn't reuse your perfectly arranged window and takes too long to open the new window.

KNode is the only newsreader I know that let's you choose a sane layout. This makes it IMO the best newsreader available. Unfortunately KMail can't do it (yet?)

Ironically Mozilla lets you choose between two perfectly useless layouts but misses the obvious. (which is folders on top-right, message overview bottom-right and message left - you get the idea, I would not mind if the message is on the right side as long as I can have it as tall as the screen)

Re:Concentrate on useful apps for 90% of people

[praedor]

Don't worry praedor, you are still on my TODO list :)

Thank GAWD! If I were a coder, or even had the time to teach myself (I am working on a thesis in molec. biology, not computer programming) I WOULD offer myself to handle it. Unfortunately, I am a plain ole enduser that needs certain functionality to do my basic day-to-day writing. The ONLY writing I do that doesn't include references these days is in emails.

Lyx IS powerful but damn, it is NOT all that user friendly. I just do not have the time right now to learn the ins and outs of latex (or all the oddball quirks of Lyx). I just want to write and move on to the next experiment.

Re:How about collaboration?

[alext]

Yes, it is a bit idealistic, fair comment.

Part of the problem, IMHO, is that C/C++ require or at least encourage the development of frameworks (GUI, RPC, storage, threading etc.) before you can get started on real functionality. This almost guarantees that development will fragment even if you start from the same place. A Java or maybe Smalltalk project would be less prone to this, though still no picnic from the convergence PoV.

Re:Some thoughts

[Ogerman]

No, again. Mozilla didn't *exist* a few years ago.

Neither did Konqueror exist in its 2.x form. (ie. when they started making it a true web browser instead of just a file manager / help viewer) You can't fairly compare 1.x Konq because it was for all practical purposes an entirely different package. I'm not saying that Mozilla is bad. It most definitely is the best browser to date. I just think the KDE folks have a superior development strategy that will propel the project faster than others with less man-hours. But that's just my opinion so take it or leave it.

Well, given that Konqueror only runs on KDE, *still* has crummy Javascript support, doesn't have as standards-compliant a renderer as Mozilla does, isn't used on the MacOS or Windows and looks and works like that god-awful IE.

Total bullshit. Moz has a mail client and a newsreader at least built in. Konqueror does not. Moz renders more HTML properly than Konqueror. As for "integrated", that's a buzzword from MS. I like app intercommunication -- drag and drop support is nice. However, tying two products is just stupid -- something designed to be done to encourage purchasing solely from one vendor.

The 3.x series is slated to entirely re-work the Javascript support, which is currently the only major hangup. Konqueror 2.2.2 is already fairly standards compliant but I believe it will become even more so in the near future.

Konqueror is not used on MacOS or Windows because it is a component of a complete desktop environment, not a standalone piece of software. So in that way, yes, it is a lot more like MS IE. On the other hand, why is that automatically a bad thing? Frankly, there are a lot of advantages to an integrated web-browser / file-manager / desktop in terms of usability. In fact, that's what Nautilus is aiming for as well, though they are way behind Konq at the moment. The average user has become very accustomed to a highly object oriented desktop environment. Sure, it's not necessary and obviously Sun hates the idea, but in terms of abstracting the user from the computer, it's a major step forward. Your comment about vendors and tying products together is entirely meaningless to Open Source projects like Konq or Moz. Either way, you could say that Moz is stupid for trying to tie a browser and mail client together. I don't use Moz's mail client so it's a waste for me. But I'm not really complaining either.