StarOffice Source Released

mprudhom writes: "According to Yahoo!, Sun has today released the source to StarOffice, as promised. Go to www.openoffice.org and download it, or just grab it with:
cvs -d:pserver:anoncvs@anoncvs.openoffice.org:/cvs login
cvs -d:pserver:anoncvs@anoncvs.openoffice.org:/cvs co OpenOffice ". Okay, people can stop submitting this now.

Comments

Small code base

[finkployd]

I would have figured the source code would have been larger than that.
And what kind of code starts with "Internal Server Error"? Is that some kind of crazy new Java function?

:)

Finkployd

Figures

[Anonymous Coward]

I was in the middle of downloading the source code, when things went to hell and the download stopped cold. I couldn't get a response from the server, so I figured I'd take a minute to see what's up at /., and what do I find, but this article.

*Sigh*

Re:The site is deader... an explanation

[pruneau]

Maybe an explanation :

Look closeley at the displayed message :

"(...) hits.
We ask your patience while our best people are reconfiguring the server and bringing her back up
(...)"

Yes, look closely : nothing strikes you ?!?!
Zoom->in : "bringing _HER_ back"

Sun just leaked yet another ground-breaking technology news : sexed servers. They choosed a female one, because they are so much caring for their users request.

But maybe it's PMS time now. Just imagine a bunch of sysadmin trying to convince the managment that boxes of tampaxes are _mandatory_ to run their web server.

Next dowtime : headaches and baby blues.

(I'still hesitating between flamebait/funny myself)

"real world" my ass...

[MenTaLguY]

Hilarious! Open Source produces another gigantic, stinking turd of a project.

Not really. This turd was formed in the bowels of proprietary software. It's only now that it's out in the open.

My personal experience with proprietary software vendors (and I've worked closely with their software engineers in some cases, trying to debug problems with their software that were creating major problems for my employer) has been that most proprietary software is complete shit, from an engineering perspective.

I mean that. There is an amazing lack of accountability for the quality of code in proprietary software.

Before I went to work in the "real world", I would never have imagined that large, well-funded companies would produce software with such egregious bugs and flawed engineering methedologies, much worse than any I have personally ever seen in any serious Open Source project (read: one with at least three active members).

Real world example (Windows), with a major EDA vendor (who will remain nameless):

• Network performance is almost unusably bad
• Vendor offers no helpful solutions
• We start looking at network traces
• It becomes plain that the software is polling all of the configured network printers in a tight loop whenever the app is active
• We go to the vendor -- it turns out to be an "architectural issue"
• We want a fix
• They suggest we don't configure any network printers

Another fun (Unix+Windows) example:

• 2d application is complaining about GLX issues
• We ask the vendor... they decided to use OpenGL instead of the 2d line-drawing primitives in Xlib (just for drawing un-transformed(!) 2d straight lines!!!)
• Inquiring further, it turns out that they don't use Xlib at all; they wrote a custom (and very slow and badly implemented) X client library instead. Which uses interrupt-driven (SIGALRM) polling for mouse events.
• Now we know why the app is so hideous to use, and why they never used any standard X widget toolkits.

I've also seen some other absolutely hair-raising things in network/system call traces, like:

seek(), ftell(), seek(), ftell(), ftell(), read(), seek()[back to same block], read()[same amount this time, but in 512k increments], seek(), ftell(), seek(), seek(), ftell(), read()

There was also the wonderful discovery that an app was using the NT equivalent of access() (GetSecurityInfo() + GetEffectiveRightsFromAcl(), which means about 40 lines of support code each time) instead of checking for failure on various operations (open file, etc) ... why?

...because the lack of error handling in the application was so pervasive, they decided to cut their losses and just anticipate all possible errors by explicitly checking for the conditions that might cause them beforehand (never mind race conditions or incomplete coverage, or the fact that it broke some things...). Things were so bad that that was actually less work and less code.

I can go on and on with these real-world accounts if you like. I've come to believe that only with Open Source comes real software engineering accountablity.

...the thing about most big projects is that they are NOT fun, NOT particularly maintainable and WELL beyond the understanding of any one coder. That's why it is necessary to PAY programmers to work (with people they might not necessarily like) IN GROUPS under the direction of others (with whom they might not necessarily agree).

Actually my experience has been that those disagrements really fuck up a software project. The Open Source projects I've been involved with, if the disagreements are really serious they usually result in a fork which often means two healthy projects rather than just one. Or the old bastard leaders are deposed and go on to other things.

Very democratic, and usually works nicely.

And the necessity of income to pay those programmers dictates that the product must be sold and that IP laws must be used to protect that income.

Only as long as you try to treat a service industry like a manufacturing industry.

And if you anyone doesn't agree with that, explain how Sun could develop an Open Source Star Office without a thriving business based on proprietary hardware.

Pretty simple: buy the rights to a proprietary product from someone else and release the source code to that. Which is what they did, actually.

(Well, they actually bought the company, as I recall, but same thing)

Now, as far as your description of what you see as the "real world", I do software support for a Fortune 500 company, and have been involved with (and contributed code to) several major Open Source projects. What experience do you have?

Fine print and caveats

[Error27]

It's funny to watch the yahoo guys try to find some way of saying that this must suck in some way.

Obviously they haven't read the GPL because it really doesn't suck at all.

In fact it's one of the greater pieces of literature of the 80's.

Some day in elementary school kids will have to memorize the GPL. And they'll have huge picture of RMS on the walls.

:P

Re:How to completely flatten a CVS server NOT

[TA]

-z3 should be perfectly ok if you're on a slow link (say, over in Europe).
The faster your link is the lower your compression setting should be. That been said, there's not much use going over -z3 though, and if you go to the highest rate you will load the server quite a bit for little or no gain. Morale: Stay at -z3 or lower, and if you're on a fast link go for -z1 or no compression.
TA

Features...

[riggwelter]

Its great that we can now take the code and add features that are sadly lacking in StarOffice, such as the MS Office Assistant...

We could have a little popup Tux penguin.

"It looks like you're writing a letter slagging off Microsoft. Would you like me to make it anonymous for you?"


--

Good for Emacs!

[kalifa]

> OpenOffice 6.05 takes a good 18 hours to compile
> on a 500Mhz win32 box, according to
> openoffice.org. Yikes :)

Yup. Thanks to the joint efforts of OpenOffice, Mozilla, and a few others, Emacs officially entered the category of lightweight utilities.

self defacing humor or self fulfilling prophecy?

[brokeninside]

On Friday the 13th -- a target fatalistically selected by Sun

Is Sun trying to say that this is something that they do not really believe in or do they just have one or more project managers with a black sense of humor?

I hope its the latter, but it would not altogether surprise me if the former was the case.

Speaking of which, does anyone see the release of StarOffice as GPL as anything other than an attempt by Sun to kill off Microsoft's cash cow, Office? Sun spent buckets and scads and tons of money on buying StarOffice, giving it away for free and then hiring CollabNet to clean up the code and modularize the CVS tree. I don't think that Sun is making enough on its SunRay thin clients to justify the expense. OTOH, a high quality, multi-platform, free Office suite might take away Microsoft's ability to subsidize W2K development with Office revenues. The question with this strategy is whether or not ms.net will be available and functional to make the desktop office suite irrelevant before a working and spectacular Star Office 6.0 for Windows is ready and available.

Hmm. Are any of the developers for this project are going to make Star Office into the free equivalent of .net? That would be funny. Hmm. Maybe the MS investment in Corel is an attempt to come up with an alternative to a free Star Office.

The most encouraging thing for me is that, to a certain extent, it seems that Sun has learned from AOL/Netscape's mistakes with Mozilla:

It has divided the massive StarOffice code base into 75 modules, grouped into 18 projects, such as printing, scripting engines, spreadsheets, and the like. Currently, all 18 projects are headed by Sun employees, but Roth said Sun is expecting "others in the community to take over some of them over time."

They made things organized and pretty and split things up into well defined sub-projects. This will make it much easier to (1) part out the useful parts of Star Office for other projects, (2) graft in new systems to fix Star Office's deficiencies, (3) keep the ball rolling, and (4) get new people involved.

Maybe after a few days and the CVS server comes back to life I'll download the code and look at it out of curiosity. I've always wondered how much of Star Office was written in Java. It's certainly slow enough at times for the whole thing to have been.

One thing is for certain, this will be an adventure....

have a day,

-l

Use compression when downloading...

[HadronPie]

cvs -d:pserver:anoncvs@anoncvs.openoffice.org:/cvs login
cvs -z3 -d:pserver:anoncvs@anoncvs.openoffice.org:/cvs co OpenOffice

Note the -z3

This will save a little on bandwidth...

6.05 First Impressions

[Anonymous Coward]

I got ahold of the Linux binary and had to try it. The good news: the download is quite a lot smaller (52.4MB, vs. something like 80.8MB for version 5.2). Evidently the web browser and mail features were taking up a lot of space. The memory footprint is also smaller: about 40MB, 30MB shared after opening several Word docs, an Excel spreadsheet, and a PowerPoint file. That horrible desktop is GONE. Separate documents open in separate windows which you can move around like they're supposed to. Maybe it's my imagination, but it seems a bit faster. Whatever toolkit-within-a-toolkit wrapper it's using still accounts for a lot of fat though. I opened a Word document that failed with 5.2, and it looks pretty damn good. Excel also. PowerPoint rendered the presentation acceptably, though the headlines were consistently kerned all bunched up together. Conclusion: didn't play with it long enough to see how much it crashes, but this seems to be headed in the right direction. A GTK version will be a killer app.

ReWrite?

[Ezz]

Projected events over the next two years: Oct 2000 - Source released Mar 2001 - Complete rewrite from ground up started Oct 2001 - JZW leaves project Jan 2002 - MS-Office2002ASP.net released Oct 2003 - OpenOffice reached build OO17 Oct 2003 - Sun release Staroffice v6 preview 3 Nov 2003 - KDE5 released with KOffice 3 Dec 2003 - OpenOffice version 1.0 released as a completely different product incorporating XML, SOAP, XUL, HTML, XHTML, Themes, Skins, Java, Bonobo, KParts, CORBA, OLE, GnomeBasic, PHP scripting and the kitchen sink.