[aur-general] aur-out-of-date: determine out-of-date AUR packages w.r.t. upstream

[Michael Kogan]

Unfortunately, I don't know Go and in general have too little programming
skills and experience to propose any code modifications. But I have an
idea: What about introducing some variables which the package maintainers
would put into their PKGBUILD and which are then parsed by aur-out-of-date?
The variables might be commented out in order to not interfere with
whatever happens when makepkg is run. Something like

# aur-out-of-date-check-url =
https://sourceforge.net/projects/sevenzip/files/7-Zip/

for the above example. There could be sane defaults defined for such
problematic hosters and only if they fail the user-defined values put into
the PKGBUILD by its maintainer would be used. Also this would require some
documentation to tell the maintainers which variables are necessary and how
to give them the desired values.

Just an idea, I don't know if it makes much sense.

2018-02-01 20:43 GMT+01:00 Simon Legner via aur-general <
aur-general at archlinux.org>:

> Hi,
>
> On Thu, Feb 1, 2018 at 5:44 PM, Eli Schwartz via aur-general
> <aur-general at archlinux.org> wrote:
> > This looks quite cool!
>
> Thanks :)
>
> > 1) How does this handle PKGBUILDs that, say, have their homepage on
> > github, but do not tag releases there -- because :( :( :( -- and instead
> > push releases to PyPI or npmjs or something?
>
> This should work: unless a version can be determined from the upstream
> URL, the first source URL is used for a second attempt. I haven't
> tested the described situation, though.
> >
> > 2) It looks like your AUR package is downloading a source tarball, then
> > completely ignoring it by using go get and downloading the bleeding-edge
> > master. I'm not really sure how this crazy language handles packaging :p
> > but I think *maybe* you should be cloning
> > git+https://github.com/simon04/$pkgname.git#tag=v$pkgver and symlinking
> > "$srcdir/$pkgname" --> "$srcdir/src/github.com/simon04/aur-out-of-date/"
>
> You're right. I hopefully fixed that in aur-out-of-date 0.6.0-2
>
> Simon
>