On Wed, Oct 19, 2011 at 4:45 AM, Ulrike Fischer <ne...@nililand.de> wrote:
> Well I'm a windows user so actually I'm not really affected. But > imho the linux distros should rethink their installation methods and > installation advices. It is absurd that 10 or more distros invest a > lot of main power in making packages when they lack the main power > to keep them up-to-date. Actually, for being a long-term support distro, Debian (using TexLive 2009) is about as up to date as you will find. Here's the reason for this. You may not agree with it but for those of us who do server programming it makes a *tremendous* amount of sense on the server side. The basic thing is that servers generally require stability because an introduced bug can affect large numbers of users simultaneously. Consequently, the way Debian does this is by running unstable versions, that graduate to testing version, which graduate to stable versions, often over a period of a couple several years. This gives early adopters an opportunity to shake out issues, and then by the time folks are deploying critical servers, the issues, limitations, etc. are well known, tested and documented, and they're not going to introduce new bugs by upgrading out from under the applications. This is important in this environment. Long term support distros (Ubuntu LTS, RHEL, Debian) tend to backport fixes for critical bugs to earlier versions where required so the software is still supported. This is one reason why which distro of TexLive is being used can be misleading. One doesn't really know what's been backported or not. This matches my needs very well. If my clients are running accounting systems, the last thing I want is an upgrade of TexLive to break their ability to generate invoices. If there are bugs in older versions, I can work around those bugs, but the problem of getting a document that will only render with one version or another is not acceptable to my application. Consequently I stick with older, solid packages, avoid cutting edge ones (exception currently being XeTeX for a subset of users, and that's only due to issues of i18n in the invoice templates, which generally causes pdflatex to croak). So this is where I am coming from. I am happy with workarounds. Not happy with "you must upgrade every couple years." Upgrades must, under no circumstances, break the accounting software, and if that means many bugs go unfixed, that's what that means. Generally speaking that means that bugs get fixed only if the maintainers conclude that the fix is backwards compatible, and that the bugfix is sufficiently non-intrusive that the chance of introducing new problems is minimal. I have already heard that this is anything but the policy of Texlive (which has other advantages, but not for the environments I work in). As a Windows user, I suspect you are thinking of desktop needs. That's fine. A lot of people use the Tex stuff as essentially desktop publishing. But there are others of us who build fairly critical systems using this and we have greatly increased needs for stability. It's one thing if a magazine, a school paper, or a book won't render because of an upgrade. It's a very different thing when a weekly batch of checks you promised your clients would be mailed out *that day* fails at 1pm in the afternoon because something changed in one of the Tex packages you use to generate the checks and now someone has to fix it in time to mail them out. The way you guarantee that is by making sure it works and not touching the underlying dependencies unless you absolutely must. The fact that they are outdated makes no difference. Best Wishes, Chris Travers -------------------------------------------------- Subscriptions, Archive, and List information, etc.: http://tug.org/mailman/listinfo/xetex
