This response is more a test case in how easy it is for a sage packaging 
novice (me) to figure out what to do, so take this with some scepticism and 
some advice from experts: I did find this one:

https://doc.sagemath.org/html/en/reference/spkg/libbraiding.html

There is obviously the "package version", but as I now see, the "equivalent 
system packages" don't seem to encode an explicit version restriction at 
all. So I guess Michael's comment on spkg-configure.m4 is probably 
appropriate. I guess explicit testing for exactly what you need there would 
be the most flexible, but the requirement there could also just be a check 
on the installed version number, I guess. In that case, sage installation 
should detect whatever system-installed libbraiding there is, run its test 
(look up the version number and compare it with the value) and then decide 
if that's good enough.

I would think it's philosophical debate on what kind of test is preferred: 
rather strict testing on version numbers would restrict you to more 
restricted but known-good configurations, whereas feature testing would be 
more permissive but risk to accept a version that ends up looking 
superficially as functional but not actually a tested-good version.

On Friday 1 November 2024 at 03:02:51 UTC-7 mmarco wrote:

> How do I fix a version requirement?
>
> El viernes, 1 de noviembre de 2024 a las 8:03:28 UTC+1, Nils Bruin 
> escribió:
>
>> On Thursday 31 October 2024 at 03:55:41 UTC-7 mmarco wrote:
>>
>> I recently released a new version of libbraiding that exposes new 
>> functionality, in order to use those new functions from Sage (see #38887). 
>>
>>
>> However, the CI tests fail because they use the old version (taken from 
>> the system, instead of installing the new spkg).
>>
>> Which is the right way to proceed in this case? Should I somehow force 
>> the containers to build the new spkg?
>>
>>
>> Would this be caught by changing the version requirement? Then the old 
>> version provided by the system wouldn't satisfy sage's version requirement 
>> any more, so wouldn't that push sage to install its own? 
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion visit 
https://groups.google.com/d/msgid/sage-devel/cf2b104d-04b8-49e0-9a34-f3a3dedca85en%40googlegroups.com.

Reply via email to