Educating packagers about always making changes in devel / rawhide first

Kevin Fenzi kevin at scrye.com
Tue Feb 11 02:31:31 UTC 2014


On Tue, 11 Feb 2014 03:04:04 +0100
Kevin Kofler <kevin.kofler at chello.at> wrote:

> Kevin Fenzi wrote:
> > This is the one I have run in the past: 
> > https://git.fedorahosted.org/cgit/releng/tree/scripts/check-upgrade-paths.py
> 
> If you're going to use that script, PLEASE use this patch:
> 
> http://lists.fedoraproject.org/pipermail/devel/attachments/20120418/ead924dd/attachment.bin

Can you attach a current version of that to a releng trac ticket and
mark it 'meeting' so it gets discussed at the next releng meeting?

That would be great. 

> I wrote, which Jesse Keating has been ignoring for YEARS despite my
> repeated nagging and complaining about the false positives in the
> nagmails.

Jesse Keating hasn't been involved in Fedora for many years. He's no
longer in any groups that could commit the above change. This is why
mailing people directly is a bad idea... please open a ticket. 

...snip...

> With my patch, you can use this path:
> 
> f19 f19-updates f19-updates-testing /f20-updates f20-updates-testing
> f21
> 
> This tells it to compare f19-updates-testing against
> f20-updates-testing rather than (or actually in addition to)
> f20-updates. (To be precise, it actually first does the
> f19-updates-testing < f20-updates check, and only if that fails, it
> redoes the comparison against f20-updates-testing.)
> 
> You can't use "/f20 /f20-updates" though, the patch is not smart
> enough to handle that. At the time, the lines used did not include GA
> releases because the check was being run after every update push, so
> checking F20 GA for broken upgrade paths would only repeat the same
> errors each time. I think that checking the f19 < f20 constraint is
> still not useful (so I didn't include it in the above enumeration),
> but f20 < f20-updates (the one constraint in my enumeration that is
> not covered by my proposed path) could be of some use. (That said, I
> do not expect that constraint to get violated in practice anyway.)

Should be easy to run and see what the output is. ;) 

kevin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/devel/attachments/20140210/0e2293b2/attachment.sig>


More information about the devel mailing list