So the problem is that if we don’t git-tag the dependencies, there is no way to reproduce a build without eups in the future, ever. I’m actually reluctant to give that up.
Right now if we git-tag, eups picks them up as versions. There is a PR open in EUPS to do the opposite, but then that means that any tags that I make (eg. the weeklies) will also get dropped.The problem is that some people care about the eups versions being consistent and some people want them to read like the initial versions and others don’t use them at all.
There is no obvious solution to this problem that keeps everybody happy. So to recap, the options are:
(a) I don’t git-tag your externals. You can never reproduce a release state without EUPS.
(b) The current situation (I git-tag your externals and the manifest shows your tag for externals - the git tags will still show you the “real” version)
(c) we switch to the new version of eups that ignores alphabetic tags, I tag your release with an alphabetic tag but then you lose the 2015_11 tag from the manifest
Open to opinions. I can call into your Wed meeting if you like.
> Hi Frossie,
>
> During the latest Qserv release, some Qserv third-party dependencies version number
> have been set to 2015_11.O. Here's the list below:
>
> (extract of https://sw.lsstcorp.org/eupspkg/tags/qserv_latest.list)
>
> xrootd generic 2015_11.0
> boost generic 2015_11.0
> protobuf generic 2015_11.0
> sconsUtils generic 2015_11.0
>
>
> Please note that for these dependencies, I think we would like to keep their "real" version
> number, so that we can easilly know which version of them we are using.
> On the other hand, I think that it is ok to use 2015_11.0 tag for packets developed by Qserv
> team (except xrootd which has its own version numbers), like you're doing now.
>
> So, for the next release, could we please switch back to "real"/previous version number
> for the 4 packages listed above? Of course, if Qserv team also agree with that.
>
> I've build a Docker image from the release you've done, and it seems fine (except that
> I had to check that xrootd 2015_11.0 was also 4.3.0.rc4a.lsst1 ;-)).
> Thanks again for packaging the release, this is a complex task and it help a lot that you
> can do it on the Square side :-)
>
> Have a nice day,
>
> Fabrice
>
--
Frossie Economou
Science Quality and Reliability Engineering (SQuaRE)
Large Synoptic Survey Telescope
Use REPLY-ALL to reply to list
To unsubscribe from the QSERV-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1