[tex-live] Stable vs. Unstable/Testing Update Repositories?

[Lars Madsen]

Zdenek Wagner wrote:
> 2010/2/23 C.M. Connelly <cmc at math.hmc.edu>:
>> I'm wondering if there could be more than one update path, with
>> (at least) one kept fairly stable (with tested, known-working
>> packages, scripts, and documentation) and one that is updated with
>> the very latest and greatest from CTAN.  It might even be worth
>> having a very stable (bugfixes-only) branch, a known-working
>> branch with the newest safe code, and a cutting-edge,
>> latest-and-greatest branch.
>>
>> I have a colleague who naively updated his TeX Live 2009 install,
>> which happily updated him to the beta version of geometry.sty,
>> which broke beamer, which he needed to compile his slides for
>> teaching later that day.
>>
> You should preferably ask the author of geometry.sty not to put beta
> versions to CTAN. The TeX Live team is not large enough to test all
> updated packages for possible problems.
> 

actually I don't think it was any fault of the geometry author

it was the beamer people who was relying on internal features in 
geometry, i.e. relying on \... at ... commands instead of official macros 
stated in the manual, that is just asking for trouble.

geometry 5 had been beta tested, but apparently the beamer people did 
not remember their test

/daleif