There have been security issues noted with the DJB DNS code. Mostly he used integers for pointers, which throws up issues on systems where the two aren't identical, not that he is the only C programmer to muddle different integer types.
But DJBs code is very tight, it also implements the bare minimum of the DNS standards, and isn't meaningfully maintained, and he made bizarre comments and requirements on the licence.
So the rest of the us use BIND 9 because it is proper free software, professionally written AND MAINTAINED, and with a perfectly respectable security record.
The real problem with BIND was version 8 and earlier, was it was written in part by Paul Vixie, who freely admits he was the source of a lot of the big security problem with early Internet code (his contributions to sendmail were also not without flaws). Then again if he hadn't written that code, chances are we'd be having this discussion on compuserv or similar.
Code from a different era with different assumptions about who'd use it. Remember when the sendmail stuff was done a lot of people left their email servers as open relays because it was more convenient for other email administrators. Oh happy days.
Never seen Fortran since I stopped being a government scientist.
Nothing wrong with the later Fortrans, but Fortran 77 is at best the 5th best language for any application (behind Fortran 90, Fortran 95, C (which is always second best allegedly), and whatever language would be the right one to write that application in.
NASA like most big government science bodies probably just has shed loads of ancient code in Fortran. When I was a govenment scientist a lot of the code said "Fortran 77" on the box, but when you peered in there was often the distinctive odour of Fortran 66, and in some cases Fortran H/G/4. You could tell by how few modern constructs the code used.
Oh and yes I got stung by some of the HP fun mentioned earlier when they messed with their fortran compiler. 'Blank common' might be the worst sort of sin in computing, but when you have many tens of thousands of lines of code using it, and other such features, it is handy if the compiler retains some backward compatibility.
So take a job where you'll learn Fortran 77 only if you favour being some backwater government scientist that maintains long dead numerical models. Refuse the job, eat bread, drink water, learn Haskell, it'll be better for your soul.
As a free software community member I don't use WINE, as it primarily exists to let people use non-free software, so my eyes aren't on it, it isn't installed on my box, and I'm not that interested in it.
Clearly there is code that has more eyes on it than others. Similarly free software sees a lot of automated code audit, which I know hasn't historically been a high priority at many commercial software organisations. But it won't spot stuff like this.
I'd guess the people interested in secure design aren't interested in reimplementing Windows APIs.
So perhaps the WINE developers missed it as well as Microsoft, but it is inevitable they will reimplement Microsoft security flaws, as so many of the problems with Windows are structural. It isn't really their job to fix Windows.
As pointed out previously though, most Windows apps still have 8.3 compatible executable names underneath the Window manager in exactly the same way GNU/Linux distros do. "Internet Explorer" is longer than 8.3, so it almost certainly isn't the real name.
Apples and Oranges, grandparent was the first post to compare like with like, and in general GNOME window manager (and KDE) both have better organised menus, and clearer names for programs than Microsoft Windows does.
The real name matters when it crashes, and the end user hits "ctrl+alt+delete", and the real names (not the menu items) are presented in some views (on both KDE and Microsoft Windows), so the hiding of real names (whichever system is involved) can create a barrier to the end user.
On my system the real name of the browser process appears to be "mozilla-firefox", so that should be pretty obvious if the end user needs to kill it. Anyone with Microsoft Windows want to compare process names?;)
Everything is grouped by it's function, unlike Windows where typically applications are grouped by manufacturer.
Where are the mod points when you need them?
Of course this isn't just Microsoft's fault, some Microsoft applications add themselves to nicely named generic menu items, like the bundled games, and some Microsoft applications don't. I blame the developers being scatters all over the world and in so many different companies, its that close source development model at work again.
-- they provide a supported, easier to configure setup which allows them to solve whatever problem they or their organization have with a minimum of fuss.
This rubs me up the wrong way. Speaking as a system admin with the choice of disto to use on our servers, and a lot of background in Redhat and other distros, I converted (am converting) them from Redhat to Debian. The only things I see in Redhat are; support, good security (at least in Fedora Core 4). In principal some of our hardware would be supported under Redhat OSes (although the one supported under 9 didn't last long). In practice we'd had one hardware support call which raised any issues concerning hardware support, and that was pretty obviously a BIOS bug, or hardware fault, eventually DELL admitted BIOS bug. But it is largely irrelevant we could have replaced the hardware for less than the cost of a copy of RHEL if push had come to shove.
It is conceivable some inexperienced admin would find Redhat installs easier, but installation is typically half a day once mostly watching data go from media to media, systems need administering long term and setting up, and Debian kicks ass here. Redhat still often feels like a bunch of applications fighting each other, in Debian you feel like the best brains have already tried to make them work together, probably because that is often the case.
Sure there are situations where you need a big enterprise support, or you want a certified Oracle platform, or you already have a bundle of these systems and just needs another "GNU/Linux" box so you stick with what you know. But assuming Debian doesn't meet people needs is naive, a lot of the ISP business is Debian everywhere. Sure there are issues with Debian (backports being the usual issue for businesses - they want all that stability and breath of support Debian stable ships in spades, but the ISPs customers often demand new and shiny as well).
Anyone who thinks Debian is for fiddling, needs to think about all those companies and organisations who have built stuff on, or from Debian, they "get it".
For us Debian is the minimum fuss option. We don't see a benefit in paying a support contract worth severals times the price of the hardware per box for running commodity type solutions in an ISP environment. If it is bust we fix it, if the fix is useful we file a bug report. In our environment Debian typically "just works", most of the bug reports and enhancement have been fringe packages, such as polishing up reporting tools.
It is shorthand for I'm a clueless buzzword user. Just look at all the IP the successful companies that support Microsft systems own, urm right virtually all of it belongs to a handful of big players, most of whom don't deliver such support services themselves but supply the software market for such services.
The reason they survive is because there is a large market for their services.
Most of the big.com companies I worked for were a few years too early, and badly managed. Interestingly these managers were generally picked for having been successful (or more likely at successful companies) before and thus appealing to the investors.
None of them had significant "IP" whatever that is, but their failures were almost all down to not getting customers onboard and paying early enough, and unrealistic expectations of marketshare (or market size) that could be achieved. In almost all cases I think the investors were foolish, but they (unlike the managers in most cases), brought with them the business contacts that could have made, or broken, the companies involved.
Whilst I like the power in KDE, the other day I failed to add a simple box to the panel showing me what applications are currently running. In GNOME this is easy, and has a lot of configurable options (apps in this workspace).
On good days I add my own regular expressions to Klipper, but in reality I'd far rather someone else sorted these things and it "just work" when I high like a URI with an odd protocol, or similar.
However I really don't think there is much to choose between these environments.
Any sane power user would just have added some "mock" printer queues with the features required already enabled underneath, if they found the GUI lacking, which would then be available from all environments (command line, KDE, GNOME, or even MACOSX or MS Windows clients).
I'm not even convinced a good Office suite. as we have come to understand it, is vital to a Desktop users success.
Microsoft Word is a letter writing application. You can just about write a structured report, such as I use to do as a consultant in it, but it becomes pretty unwieldy when you try and number paragraphs, and do tables of contents, and other basic report writing stuff. If you try and insert boxes, diagrams and the like, Microsoft Word quickly becomes totally unusable, even to the point of crashing, bailing you out into macro source code, and other bizarre activities. Word isn't even terribly good at letter writing, as I've seen very experienced Office users fail to print out labels in Microsoft Office, do the same task in OpenOffice (1) in a few seconds, because someone obviously worked through this as a "Use case" of an Office suite.
Okay some people do manage to write books in Word, but my experience is most professional writers (especially technical writers) are using applications designed to support them properly.
Similarly I've only seen two cases where spreadsheets were used in any meaningful way in a long career as an IT consultant. One was for bidding/pricing contracts, and the other was a pivot table system for reporting from databases. In almost all cases what I see people using spreadsheets for would be better done in a trivial database application, and probably easier to do. Quite a lot of people use Excel as a graphing tool, which use to be a third party plugin from someone else anyway (and a very good graphing tool it is).
Presentation software is probably the application I see used most often for the obviously designed for purpose. However I've noticed a lot of people don't even use Powerpoint. There is a whole range of software for presentations, and these days people often use stuff that integrates better with their websites, which is another simple way of embedding multimedia for people (although not necessarily clever, I've already been to presentations where lack of a fast Internet connection meant the multimedia bits weren't available), and produces a better Internet version of the same material that lets the bigger audience of non-present people read the same material later (since presentations are nearly always selling of some sort, even if only of an idea).
That isn't to say that "not running MS Office" isn't a problem in the market for Linux, but I use OpenOffice primarily as a letter writing application, and as a Word document viewer to read data from organisations too narrow minded to realise not everyone wants data in closed formats.
What I actually want on both platforms is a decent letter writing application. At work most of the business letters are printed from the database direct by little Perl scripts, automation of this type is suitable for all but the smallest businesses.
I write most of my documents in ASCII text using a text editor, and the ready ability to search and find stuff, cut and paste without formatting issues, and the like makes me far more productive than I find any of the Word Processors. Although these text editors are essentially word processors, just without messing around with features that get in the way like fonts, italics, and the like.
So yes Microsoft Word has a few features extra. OpenOffice is I think easier to use, but less featured and slower. But my day to day computing experience is browser, mail client, and specialist applications. I suspect organisation where Office suites are a major components of day to day computing are doing computing wrong, with the possible exception of sales and management roles. Although some companies have built their automation of letter writing and the like around Microsoft Office, which is an okay strategy, but tends to hurt on the upgrade cycles. The smart bunnies use simpler more open technology to do this sort of thing, and can always chuck the results into the latest Office suite if it needs "post processing".
Even though SCO they have a snowballs chance in hell of winning anything worth having, it is in the interests of people with deep pockets to keep the case going as long as possible.
There is enough at stake that for others it is worth a punt of a few million, just on the rare off chance they might get a settlement worth having a share of.
As such I'd be surprised if some entity wasn't around for the court case. Whether there is anyone else there but lawyers by that point is another question entirely. The way the system works these days, it is likely anything worth preserving will be sold off.
Are there restrictions on SCO divesting itself of assets (especially at a good price - know what I mean), as I doubt IBM wants to win, and then find there isn't enough left to cover their costs?
My experience is Ubuntu is Debian done faster, less carefully (I doubt they fix all the release critical bugs), and less extensively (platforms and packages).
That might make it fine for a Desktop OS, but my experiences with it make me very wary of using it on servers.
Given the interviewer hadn't quite got the "job" title right, perhaps he hadn't heard of Ubuntu;)
But I don't think Branden is going to have any interesting insights on Ubuntu. The interesting insights I think come from those doing Ubuntu.
Ubuntu only changes Debian in terms of what it takes away in users/developers, and even then most of it is reusable in Debian easily enough. What I want to know is what the people who "left" Debian development for Ubuntu development think and do.
Freeduc is cool. I gave a demo to local LUG, and invited loads of teachers along (a few even came, in fact we hardly recognised any of the regulars in the crowd), gave out CDs, early 2003.
I even learnt stuff about the shepherding moons of Saturn whilst preparing my talk, some of the astronomy apps were really astounding (even if it did keep trying to predict the location of Mir and getting it way off!), it was even more amazing they were all piled on one LiveCD with the OS, Wordprocessor, Spreadsheet, and bucket loads of other software, and support for seven languages at the time.
Freeduc also has LTSP integrated. And Apache starts up to demonstrate some web based classroom software, including a very good maths quiz system with integrated reporting so the teachers could know who had done what exercises etc.
Of course this is where I first used "tuxpaint", which is an essential app for all distros aimed at small children (and the young at heart).
However I don't think my talk led to too many new converts to "GNU/Linux" desktops in schools. As such I don't think the success of GNU/Linux is really dependent on availability of distros, or LiveCDs.
Freeduc really demonstrated how much could be done for so little in the Educational sector, if only people would exercise some imagination. But how you sell this to the people that matter I don't know, take them for drinks on expensive yachts in warm climes I suspect.
For schools this is better done at the network level if it needs to be done at all.
Squid has been used extensively by schools as a proxy, and the Squid community has a number of projects to do this sort of thing.
I think home use is a harder thing to enforce (you could use similar tools if you knew how to), but I've never seen any tools do a good job at this.
Not least I think those who want to protect our children from certain types of contents often are just projecting there own hangups, and prejudices.
As regards explicit content, kids really aren't that interested until the hormones start raging, and they'll soon find some sort of outlet for these feelings at that point, computers probably being the least of a parent worries (computer don't get pregnant).
> Real systems like Champion controller and sage and Cougar mountian or even Excalibur.
Oracle Financials
I mean if your customisation bill for an upgrade to the accouting system isn't over a quarter of a million, it is a mickey mouse accounting package. Well thats what the Oracle sales guy implied;)
It's just accounting, give me something that records where the money came from and went to in a sensible database, and a good IT department can do the rest. Really people go overboard on "solutions" which add very little value generally. I suspect they only get bought because it appears, when they are sold, to save work for the beancounters who sign the cheques for them.
> Think how attractive it would be if RH / Novell could back up and install over an NT server.
What would be the point? Much as I'm loathed to work with Windows servers, once it is in and installed and working it is delivering benefit. It is only when the current architecture is giving pain, or needs replacing through age anyway, that big corporate will roll out new systems (except for financial institution with more money than sense who seem to roll out whatever is flavour of the month, sometime almost that often).
I mean if I was deploying Linux I'd expect to reduce the server count of most big NT sites by a factor of 2 to 10 if not more.
Do you think Novell don't have tools for migrating the directory service, which is probably the useful bit?
The person says they have no linux experience, well heck if he had 40,000 desktops to roll out he could probably afford to hire some to make sure it got done right, so that isn't really a useful answer.
None of their applications work under Linux - right. and no doubt back in 1993 Windows ran all their apps when they switched from SCO|OS400|CP/M|MVS|VMS|Unix (delete as applicable).
Add to my list of people too hire (currently someone who has done it before, and someone with skilled lowlevel Linux skills) a couple of people who remember computing before Windows dominated the desktop.
For read-only stuff (i.e. network application locations), NFS will do redundancy fine in association with automounting, been there, done that, with HP-UX.
Have to say whilst I've managed several moderate sized (think 10 to 100 desktops) groups of Unix desktops (mostly HP-UX), I've never managed really big desktop roll-outs (100's/1000's). I did work with a group looking after 100's of SUN workstations, but that was small groups (often only 1 or 2) of workstations per site scattered all over Europe, and some of that was done with a lot of leg work (I know, my legs) but that was because bandwidth for private networking was expensive at the time, so remote installs of isolated boxes for more than a few packages unworkable, although we did miracles with what bandwidth was available.
The only people I know who have done that size in anger, in a "typical" (well fairly atypical but you know what I mean) corporate environment, and for long periods were SUN. They had Solaris everywhere. They had a corporate build policy out of HQ, which specified almost everything from the partition sizes for particular disks up. The boxes were all 'Jump started', so a network install from booting. Local admins managed the 'Jump start' configuration particulars. We got some training off one of their inhouse admins, he was surprisingly unbusy considering the number of desktops he was responsible for, but he was VERY knowledgable. I think directory services were from local NIS servers, but that was back when SUN were converting the world to NIS+ (now there is a technology that never happened).
The reason why LTSP is discussed everywhere, is because for most businesses doing really big Unix or Linux roll-out there is no business case for desktops with as many configurable components as Windows gives you. So the cost justification for the switch is instead of having 40,000 machines to manage, you have maybe 4000 and a lot of "dumbish" terminals, and manage it with a tenth of the staff/effort. Although I doubt the staff savings are that drastic, as there is usually a requirement to have feet on the ground in any moderately sized office, but they can be doing something more useful than swatting virus outbreaks, or pondering the lastest weird instability that Windows has developed.
The main exception, and the historical home of big Unix desktop roll outs (with thicker clients) was CAD, and related engineering tasks, and even there I saw installations with too many (complex) workstations, and too few servers and thin clients (in one case they really needed a supercomputer, not N high powered HP workstations).
I don't think there is a "right way", looking for one perhaps misses the point, some people buy into expensive support tools from HP that queue updates, and run dry installs and report back (good for servers, not so obviously useful for client systems). Others roll their own out of cfengine, or just reinstall the clients from scratch for updates(100Mbps networking brought some new ideas). Everyone reads the SUN "Big Admin" site avidly, because they know.
I've also worked with larger green screen roll-outs, but they are out of fashion these days, despite them always having been the least troublesome. You delivered apps to thousands of people, and you never gave the "end desktop" a thought (except if they printed through VT escape characters -- yuk don't do it -- but printing is always troublesome). The current equivalent would probably be web applications.
The one thing you will need is admins who have done it before, although not necessarily with Linux, since Unix skills transfer well in this environment.
And also people who have real knowledge of the low level stuff happening in the network installs (was OpenBoot or HP firmware, bootp, tftp, X, and the OS boot procedure but is likely to be Wake-on-LAN, PXE, dhcp, tftp, X (or VNC), and the OS boot procedure these days -- oh how quickly things have changed).
>... any kind of results are possible which DOES NOT mean that the poll was biased or flawed....
Yes, but when the statistics throw up something as dramatic as more Desktop Ubuntu than Redhat, you question the representativeness of the statistics, because it is extremely unlikely that would occur by chance.
Obvious browser useragents should include the entire state of the OS, and we could settle this for the webbrowsing community. I looked in my referer log, it is either Linux or Debian, don't see none of the Ubuntu, or Redhat stuff;-). Our weblogs suggest rather more browsers claiming to be Debian than Ubuntu, but that could be Ubuntu claiming to be Debian.
On the other hand we have one server, with Debian, and two virtual Debian instances running on it, so a quick IP address scan will show up most of our servers as GNU/Linux from this, and one other GNU/Linux box with multiple IP addresses, when in fact the balance is closer to 50:50.
Of course we bought this server secondhand so it won't show up in the statistics, except it was running NT4 before, so somehow that went from being one server on the Microsoft side, to being three servers on the GNU/Linux side, with no one telling Gartner Inc.
How many were sold without OS? I'm guessing the revenue from those is a lot less, since most of our server hardware is cheaper than the corresponding software licences (RHEL or W2K3).
Re:Not 1-man distros so much as derivitive distros
on
Libranet On The Rocks
·
· Score: 1
> which means Ubuntu is pointless.
I'm no great Ubuntu fan, but I don't see why that merely because it is a Debian derivative, and would have a hard time if Debian folded it is pointless. That logic would make all Debian derivatives pointless, and some of them are definitely useful.
Sure if people buy into Ubuntu because it has a more business friendly footing, under the impression its future is somehow more robust than Debians because of that (and nothing to do with Mark's pockets), they may be wrong.
I do wonder if Ubuntu would be better focused entirely on the Desktop, although no doubt there is a certain need for servers that are compatible with their clients.
> About those commandline shortcuts, I'd say that these should be disabled by default.
Please no - they are there for a reason.
They are on obscure key combinations for a reason.
The distros should document how to use the "compose" key to create other common European characters, and make sure "compose" is mapped to a key "out of the box". Heck composing foreign character is much easier under X than it ever was under MS Windows, I don't understand why Microsoft didn't just copy it, since I doubt anyone would seriously have objected.
Sure "ctrl-alt" + "random keys" might cause the odd funny thing to happen, but it does that in Windows as well. My current Linux desktop just disabled these shortcuts at the GDM login screen, and as a system admin you're completely stuck if you can't get underneath the GUI and the GUI fouls up. I mean if we wanted an OS that was toast as soon as the GUI messes up we'd run MS Windows.
Slashdot Mirror
There have been security issues noted with the DJB DNS code. Mostly he used integers for pointers, which throws up issues on systems where the two aren't identical, not that he is the only C programmer to muddle different integer types.
But DJBs code is very tight, it also implements the bare minimum of the DNS standards, and isn't meaningfully maintained, and he made bizarre comments and requirements on the licence.
So the rest of the us use BIND 9 because it is proper free software, professionally written AND MAINTAINED, and with a perfectly respectable security record.
The real problem with BIND was version 8 and earlier, was it was written in part by Paul Vixie, who freely admits he was the source of a lot of the big security problem with early Internet code (his contributions to sendmail were also not without flaws). Then again if he hadn't written that code, chances are we'd be having this discussion on compuserv or similar.
Code from a different era with different assumptions about who'd use it. Remember when the sendmail stuff was done a lot of people left their email servers as open relays because it was more convenient for other email administrators. Oh happy days.
Never seen Fortran since I stopped being a government scientist.
Nothing wrong with the later Fortrans, but Fortran 77 is at best the 5th best language for any application (behind Fortran 90, Fortran 95, C (which is always second best allegedly), and whatever language would be the right one to write that application in.
NASA like most big government science bodies probably just has shed loads of ancient code in Fortran. When I was a govenment scientist a lot of the code said "Fortran 77" on the box, but when you peered in there was often the distinctive odour of Fortran 66, and in some cases Fortran H/G/4. You could tell by how few modern constructs the code used.
Oh and yes I got stung by some of the HP fun mentioned earlier when they messed with their fortran compiler. 'Blank common' might be the worst sort of sin in computing, but when you have many tens of thousands of lines of code using it, and other such features, it is handy if the compiler retains some backward compatibility.
So take a job where you'll learn Fortran 77 only if you favour being some backwater government scientist that maintains long dead numerical models. Refuse the job, eat bread, drink water, learn Haskell, it'll be better for your soul.
As a free software community member I don't use WINE, as it primarily exists to let people use non-free software, so my eyes aren't on it, it isn't installed on my box, and I'm not that interested in it.
Clearly there is code that has more eyes on it than others. Similarly free software sees a lot of automated code audit, which I know hasn't historically been a high priority at many commercial software organisations. But it won't spot stuff like this.
I'd guess the people interested in secure design aren't interested in reimplementing Windows APIs.
So perhaps the WINE developers missed it as well as Microsoft, but it is inevitable they will reimplement Microsoft security flaws, as so many of the problems with Windows are structural. It isn't really their job to fix Windows.
As pointed out previously though, most Windows apps still have 8.3 compatible executable names underneath the Window manager in exactly the same way GNU/Linux distros do. "Internet Explorer" is longer than 8.3, so it almost certainly isn't the real name.
;)
Apples and Oranges, grandparent was the first post to compare like with like, and in general GNOME window manager (and KDE) both have better organised menus, and clearer names for programs than Microsoft Windows does.
The real name matters when it crashes, and the end user hits "ctrl+alt+delete", and the real names (not the menu items) are presented in some views (on both KDE and Microsoft Windows), so the hiding of real names (whichever system is involved) can create a barrier to the end user.
On my system the real name of the browser process appears to be "mozilla-firefox", so that should be pretty obvious if the end user needs to kill it. Anyone with Microsoft Windows want to compare process names?
Where are the mod points when you need them?
Of course this isn't just Microsoft's fault, some Microsoft applications add themselves to nicely named generic menu items, like the bundled games, and some Microsoft applications don't. I blame the developers being scatters all over the world and in so many different companies, its that close source development model at work again.
This rubs me up the wrong way. Speaking as a system admin with the choice of disto to use on our servers, and a lot of background in Redhat and other distros, I converted (am converting) them from Redhat to Debian. The only things I see in Redhat are; support, good security (at least in Fedora Core 4). In principal some of our hardware would be supported under Redhat OSes (although the one supported under 9 didn't last long). In practice we'd had one hardware support call which raised any issues concerning hardware support, and that was pretty obviously a BIOS bug, or hardware fault, eventually DELL admitted BIOS bug. But it is largely irrelevant we could have replaced the hardware for less than the cost of a copy of RHEL if push had come to shove.
It is conceivable some inexperienced admin would find Redhat installs easier, but installation is typically half a day once mostly watching data go from media to media, systems need administering long term and setting up, and Debian kicks ass here. Redhat still often feels like a bunch of applications fighting each other, in Debian you feel like the best brains have already tried to make them work together, probably because that is often the case.
Sure there are situations where you need a big enterprise support, or you want a certified Oracle platform, or you already have a bundle of these systems and just needs another "GNU/Linux" box so you stick with what you know. But assuming Debian doesn't meet people needs is naive, a lot of the ISP business is Debian everywhere. Sure there are issues with Debian (backports being the usual issue for businesses - they want all that stability and breath of support Debian stable ships in spades, but the ISPs customers often demand new and shiny as well).
Anyone who thinks Debian is for fiddling, needs to think about all those companies and organisations who have built stuff on, or from Debian, they "get it".
For us Debian is the minimum fuss option. We don't see a benefit in paying a support contract worth severals times the price of the hardware per box for running commodity type solutions in an ISP environment. If it is bust we fix it, if the fix is useful we file a bug report. In our environment Debian typically "just works", most of the bug reports and enhancement have been fringe packages, such as polishing up reporting tools.
I use KDE at work, and GNOME at home.
Whilst I like the power in KDE, the other day I failed to add a simple box to the panel showing me what applications are currently running. In GNOME this is easy, and has a lot of configurable options (apps in this workspace).
On good days I add my own regular expressions to Klipper, but in reality I'd far rather someone else sorted these things and it "just work" when I high like a URI with an odd protocol, or similar.
However I really don't think there is much to choose between these environments.
Any sane power user would just have added some "mock" printer queues with the features required already enabled underneath, if they found the GUI lacking, which would then be available from all environments (command line, KDE, GNOME, or even MACOSX or MS Windows clients).
Did they remember to flush the shell history file between tests?
I'm not even convinced a good Office suite. as we have come to understand it, is vital to a Desktop users success.
Microsoft Word is a letter writing application. You can just about write a structured report, such as I use to do as a consultant in it, but it becomes pretty unwieldy when you try and number paragraphs, and do tables of contents, and other basic report writing stuff. If you try and insert boxes, diagrams and the like, Microsoft Word quickly becomes totally unusable, even to the point of crashing, bailing you out into macro source code, and other bizarre activities. Word isn't even terribly good at letter writing, as I've seen very experienced Office users fail to print out labels in Microsoft Office, do the same task in OpenOffice (1) in a few seconds, because someone obviously worked through this as a "Use case" of an Office suite.
Okay some people do manage to write books in Word, but my experience is most professional writers (especially technical writers) are using applications designed to support them properly.
Similarly I've only seen two cases where spreadsheets were used in any meaningful way in a long career as an IT consultant. One was for bidding/pricing contracts, and the other was a pivot table system for reporting from databases. In almost all cases what I see people using spreadsheets for would be better done in a trivial database application, and probably easier to do. Quite a lot of people use Excel as a graphing tool, which use to be a third party plugin from someone else anyway (and a very good graphing tool it is).
Presentation software is probably the application I see used most often for the obviously designed for purpose. However I've noticed a lot of people don't even use Powerpoint. There is a whole range of software for presentations, and these days people often use stuff that integrates better with their websites, which is another simple way of embedding multimedia for people (although not necessarily clever, I've already been to presentations where lack of a fast Internet connection meant the multimedia bits weren't available), and produces a better Internet version of the same material that lets the bigger audience of non-present people read the same material later (since presentations are nearly always selling of some sort, even if only of an idea).
That isn't to say that "not running MS Office" isn't a problem in the market for Linux, but I use OpenOffice primarily as a letter writing application, and as a Word document viewer to read data from organisations too narrow minded to realise not everyone wants data in closed formats.
What I actually want on both platforms is a decent letter writing application. At work most of the business letters are printed from the database direct by little Perl scripts, automation of this type is suitable for all but the smallest businesses.
I write most of my documents in ASCII text using a text editor, and the ready ability to search and find stuff, cut and paste without formatting issues, and the like makes me far more productive than I find any of the Word Processors. Although these text editors are essentially word processors, just without messing around with features that get in the way like fonts, italics, and the like.
So yes Microsoft Word has a few features extra. OpenOffice is I think easier to use, but less featured and slower. But my day to day computing experience is browser, mail client, and specialist applications. I suspect organisation where Office suites are a major components of day to day computing are doing computing wrong, with the possible exception of sales and management roles. Although some companies have built their automation of letter writing and the like around Microsoft Office, which is an okay strategy, but tends to hurt on the upgrade cycles. The smart bunnies use simpler more open technology to do this sort of thing, and can always chuck the results into the latest Office suite if it needs "post processing".
Even though SCO they have a snowballs chance in hell of winning anything worth having, it is in the interests of people with deep pockets to keep the case going as long as possible.
There is enough at stake that for others it is worth a punt of a few million, just on the rare off chance they might get a settlement worth having a share of.
As such I'd be surprised if some entity wasn't around for the court case. Whether there is anyone else there but lawyers by that point is another question entirely. The way the system works these days, it is likely anything worth preserving will be sold off.
Are there restrictions on SCO divesting itself of assets (especially at a good price - know what I mean), as I doubt IBM wants to win, and then find there isn't enough left to cover their costs?
I really wish people would stop calling debian (and even linux) a product.
I'll call Debian a "solution" if it'll make you happy ;)
My experience is Ubuntu is Debian done faster, less carefully (I doubt they fix all the release critical bugs), and less extensively (platforms and packages).
That might make it fine for a Desktop OS, but my experiences with it make me very wary of using it on servers.
Faster I like....
Given the interviewer hadn't quite got the "job" title right, perhaps he hadn't heard of Ubuntu ;)
But I don't think Branden is going to have any interesting insights on Ubuntu. The interesting insights I think come from those doing Ubuntu.
Ubuntu only changes Debian in terms of what it takes away in users/developers, and even then most of it is reusable in Debian easily enough. What I want to know is what the people who "left" Debian development for Ubuntu development think and do.
Freeduc is cool. I gave a demo to local LUG, and invited loads of teachers along (a few even came, in fact we hardly recognised any of the regulars in the crowd), gave out CDs, early 2003.
I even learnt stuff about the shepherding moons of Saturn whilst preparing my talk, some of the astronomy apps were really astounding (even if it did keep trying to predict the location of Mir and getting it way off!), it was even more amazing they were all piled on one LiveCD with the OS, Wordprocessor, Spreadsheet, and bucket loads of other software, and support for seven languages at the time.
Freeduc also has LTSP integrated. And Apache starts up to demonstrate some web based classroom software, including a very good maths quiz system with integrated reporting so the teachers could know who had done what exercises etc.
Of course this is where I first used "tuxpaint", which is an essential app for all distros aimed at small children (and the young at heart).
However I don't think my talk led to too many new converts to "GNU/Linux" desktops in schools. As such I don't think the success of GNU/Linux is really dependent on availability of distros, or LiveCDs.
Freeduc really demonstrated how much could be done for so little in the Educational sector, if only people would exercise some imagination. But how you sell this to the people that matter I don't know, take them for drinks on expensive yachts in warm climes I suspect.
For schools this is better done at the network level if it needs to be done at all.
Squid has been used extensively by schools as a proxy, and the Squid community has a number of projects to do this sort of thing.
I think home use is a harder thing to enforce (you could use similar tools if you knew how to), but I've never seen any tools do a good job at this.
Not least I think those who want to protect our children from certain types of contents often are just projecting there own hangups, and prejudices.
As regards explicit content, kids really aren't that interested until the hormones start raging, and they'll soon find some sort of outlet for these feelings at that point, computers probably being the least of a parent worries (computer don't get pregnant).
It doesn't let you surf ".xxx" domain names ;)
> Real systems like Champion controller and sage and Cougar mountian or even Excalibur.
;)
Oracle Financials
I mean if your customisation bill for an upgrade to the accouting system isn't over a quarter of a million, it is a mickey mouse accounting package. Well thats what the Oracle sales guy implied
It's just accounting, give me something that records where the money came from and went to in a sensible database, and a good IT department can do the rest. Really people go overboard on "solutions" which add very little value generally. I suspect they only get bought because it appears, when they are sold, to save work for the beancounters who sign the cheques for them.
> Think how attractive it would be if RH / Novell could back up and install over an NT server.
What would be the point? Much as I'm loathed to work with Windows servers, once it is in and installed and working it is delivering benefit. It is only when the current architecture is giving pain, or needs replacing through age anyway, that big corporate will roll out new systems (except for financial institution with more money than sense who seem to roll out whatever is flavour of the month, sometime almost that often).
I mean if I was deploying Linux I'd expect to reduce the server count of most big NT sites by a factor of 2 to 10 if not more.
Do you think Novell don't have tools for migrating the directory service, which is probably the useful bit?
Why isn't the original response modded troll?
The person says they have no linux experience, well heck if he had 40,000 desktops to roll out he could probably afford to hire some to make sure it got done right, so that isn't really a useful answer.
None of their applications work under Linux - right. and no doubt back in 1993 Windows ran all their apps when they switched from SCO|OS400|CP/M|MVS|VMS|Unix (delete as applicable).
Add to my list of people too hire (currently someone who has done it before, and someone with skilled lowlevel Linux skills) a couple of people who remember computing before Windows dominated the desktop.
For read-only stuff (i.e. network application locations), NFS will do redundancy fine in association with automounting, been there, done that, with HP-UX.
Have to say whilst I've managed several moderate sized (think 10 to 100 desktops) groups of Unix desktops (mostly HP-UX), I've never managed really big desktop roll-outs (100's/1000's). I did work with a group looking after 100's of SUN workstations, but that was small groups (often only 1 or 2) of workstations per site scattered all over Europe, and some of that was done with a lot of leg work (I know, my legs) but that was because bandwidth for private networking was expensive at the time, so remote installs of isolated boxes for more than a few packages unworkable, although we did miracles with what bandwidth was available.
The only people I know who have done that size in anger, in a "typical" (well fairly atypical but you know what I mean) corporate environment, and for long periods were SUN. They had Solaris everywhere. They had a corporate build policy out of HQ, which specified almost everything from the partition sizes for particular disks up. The boxes were all 'Jump started', so a network install from booting. Local admins managed the 'Jump start' configuration particulars. We got some training off one of their inhouse admins, he was surprisingly unbusy considering the number of desktops he was responsible for, but he was VERY knowledgable. I think directory services were from local NIS servers, but that was back when SUN were converting the world to NIS+ (now there is a technology that never happened).
The reason why LTSP is discussed everywhere, is because for most businesses doing really big Unix or Linux roll-out there is no business case for desktops with as many configurable components as Windows gives you. So the cost justification for the switch is instead of having 40,000 machines to manage, you have maybe 4000 and a lot of "dumbish" terminals, and manage it with a tenth of the staff/effort. Although I doubt the staff savings are that drastic, as there is usually a requirement to have feet on the ground in any moderately sized office, but they can be doing something more useful than swatting virus outbreaks, or pondering the lastest weird instability that Windows has developed.
The main exception, and the historical home of big Unix desktop roll outs (with thicker clients) was CAD, and related engineering tasks, and even there I saw installations with too many (complex) workstations, and too few servers and thin clients (in one case they really needed a supercomputer, not N high powered HP workstations).
I don't think there is a "right way", looking for one perhaps misses the point, some people buy into expensive support tools from HP that queue updates, and run dry installs and report back (good for servers, not so obviously useful for client systems). Others roll their own out of cfengine, or just reinstall the clients from scratch for updates(100Mbps networking brought some new ideas). Everyone reads the SUN "Big Admin" site avidly, because they know.
I've also worked with larger green screen roll-outs, but they are out of fashion these days, despite them always having been the least troublesome. You delivered apps to thousands of people, and you never gave the "end desktop" a thought (except if they printed through VT escape characters -- yuk don't do it -- but printing is always troublesome). The current equivalent would probably be web applications.
The one thing you will need is admins who have done it before, although not necessarily with Linux, since Unix skills transfer well in this environment.
And also people who have real knowledge of the low level stuff happening in the network installs (was OpenBoot or HP firmware, bootp, tftp, X, and the OS boot procedure but is likely to be Wake-on-LAN, PXE, dhcp, tftp, X (or VNC), and the OS boot procedure these days -- oh how quickly things have changed).
>... any kind of results are possible which DOES NOT mean that the poll was biased or flawed....
;-). Our weblogs suggest rather more browsers claiming to be Debian than Ubuntu, but that could be Ubuntu claiming to be Debian.
Yes, but when the statistics throw up something as dramatic as more Desktop Ubuntu than Redhat, you question the representativeness of the statistics, because it is extremely unlikely that would occur by chance.
Obvious browser useragents should include the entire state of the OS, and we could settle this for the webbrowsing community. I looked in my referer log, it is either Linux or Debian, don't see none of the Ubuntu, or Redhat stuff
On the other hand we have one server, with Debian, and two virtual Debian instances running on it, so a quick IP address scan will show up most of our servers as GNU/Linux from this, and one other GNU/Linux box with multiple IP addresses, when in fact the balance is closer to 50:50.
Of course we bought this server secondhand so it won't show up in the statistics, except it was running NT4 before, so somehow that went from being one server on the Microsoft side, to being three servers on the GNU/Linux side, with no one telling Gartner Inc.
How many were sold without OS? I'm guessing the revenue from those is a lot less, since most of our server hardware is cheaper than the corresponding software licences (RHEL or W2K3).
> which means Ubuntu is pointless.
I'm no great Ubuntu fan, but I don't see why that merely because it is a Debian derivative, and would have a hard time if Debian folded it is pointless. That logic would make all Debian derivatives pointless, and some of them are definitely useful.
Sure if people buy into Ubuntu because it has a more business friendly footing, under the impression its future is somehow more robust than Debians because of that (and nothing to do with Mark's pockets), they may be wrong.
I do wonder if Ubuntu would be better focused entirely on the Desktop, although no doubt there is a certain need for servers that are compatible with their clients.
> About those commandline shortcuts, I'd say that these should be disabled by default.
Please no - they are there for a reason.
They are on obscure key combinations for a reason.
The distros should document how to use the "compose" key to create other common European characters, and make sure "compose" is mapped to a key "out of the box". Heck composing foreign character is much easier under X than it ever was under MS Windows, I don't understand why Microsoft didn't just copy it, since I doubt anyone would seriously have objected.
Sure "ctrl-alt" + "random keys" might cause the odd funny thing to happen, but it does that in Windows as well. My current Linux desktop just disabled these shortcuts at the GDM login screen, and as a system admin you're completely stuck if you can't get underneath the GUI and the GUI fouls up. I mean if we wanted an OS that was toast as soon as the GUI messes up we'd run MS Windows.