[MacPorts] #26763: sane-frontends: fix livecheck
MacPorts
noreply at macports.org
Tue Oct 12 11:44:59 PDT 2010
#26763: sane-frontends: fix livecheck
----------------------------------------------+-----------------------------
Reporter: aguynamedryan+macports@… | Owner: macports-tickets@…
Type: defect | Status: new
Priority: Normal | Milestone:
Component: ports | Version: 1.9.1
Keywords: haspatch maintainer | Port: sane-frontends
----------------------------------------------+-----------------------------
Comment(by aguynamedryan+macports@…):
You are correct that a hash is a poor choice for a version number, as I
learned in my port of pianobar.
However, the Portfile for sane-frontends currently lists the version as
1.0.15git which is the version output by {{{scanadf --version}}} There
was much discussion about this in #25267 when I submitted the original
Portfile.
It is my impression that until sane-frontends 1.0.15 is officially
released, the SANE developers will continue to use 1.0.15git for all
future updates to sane-frontends. It also appears that 1.0.15 might take
a while to be released. 1.0.14 was released over 5 years ago. My plan
was to track the master branch of the sane-frontends git repo and release
a new revision to the sane-frontends port each time a new commit is made.
New commits seem to be a rare event. In the last 4.5 years there have
been 16 commits and only 1 commit in the last year.
Instead of resorting to an epoch, which I agree is heavy handed, couldn't
we use a revision number in the Portfile to track updates I make to the
port?
Hopefully that addresses the concerns about the sane-frontends version
number. On to the reason for my patch: livecheck
While hashes aren't awesome for version numbers, they are good for
tracking git commits. The patch I'm submitting deals with livecheck-ing
the sane-frontends git repo. For the livecheck, I only want to know when
the master branch has changed. I assume that a change to the master
branch implies there is an update to sane-frontends that I need to
investigate. The Portfile already has git.branch set for the fetch phase.
It made sense to me to use that hash to compare against the sane-frontends
git repo as opposed to storing the date of the last commit or something
like that in the Portfile.
I've looked through most of the other ports that are using git.branch
somewhere in their Portfile and my livecheck approach does seem
unorthodox, but I didn't come across any solutions that fit my problem.
Most other solutions relied on the upstream developers using tags or
releasing new versions. I'm open to suggestions on better ways to do the
livecheck in this situation where the SANE developers don't appear to be
releasing or tagging new versions of sane-frontends.
--
Ticket URL: <https://trac.macports.org/ticket/26763#comment:3>
MacPorts <http://www.macports.org/>
Ports system for Mac OS
More information about the macports-tickets
mailing list