[Piuparts-devel] piuparts server packaging

[Dave Steele]

On Sun, Jun 17, 2012 at 2:05 PM, Andreas Beckmann <debian at abeckmann.de> wrote:
> What belongs to which package?
>
> Who puts something into /etc?

I'm guessing the older version of piuparts. It was a meta-failure on
the piuparts package.

>
> How does installation via update-piuparts-setup interfere with package
> installation?
>

I don't know, I don't use it. I suspect that it will work ok as long
as you keep up with the directory changes.

> I'd like to ship scripts*/ together with piuparts as these scripts are
> useful for standalone piuparts runs, too.
>
>
> 3c0649bb5:
> this does not look correct:
>  dh_installman --sourcedir=debian/piuparts
>

I had a reason for it, but that code is from last year - I don't
remember the specifics.

BTW, I tried relative addressing on the ht_root work again. None of
the links were successful. I think I knew that last year too.

> bc69493aa:
> readme-server.txt references /usr/sbin/slave_* without piuparts_ prefix
>
> postinst:
> use homedirectory /var/lib/piuparts/$user
> s/grep >/dev/null/grep -q/
> don't use $user after the loop, just use $suser
> chown <user><COLON_not_full_stop><group>
> WTF: rm -rf /var/lib/piuparts/htdocs
>  on every postinst configure?

It saves you from having to consider how other versions of the
software laid it out.

>
> postrm looks useless

Seems so, but I had a failure without it.

>
> there are changes to update-piuparts-setup that seem incorrect

I changed all references to the old piuparts* home directories.

>
> wasn't there a reason to keep the python modules in the piuparts
> package? especially if we decide to split out slave later ...
>
> 5cf23a1f2ea:
> I don't agree to serving docs from piatti, the should come from the
> local instance (but there could be a link to the latest version on piatti)
>

Your comment seems contradictory, but I agree. It was a short-term fix
to avoid having to massage the docs, or have broken links. I assume
that piuparts.d.o will continue to have a setup script in the short
term.

> the doc_root needs cleanup and the parameter needs sanitized: a trailing
> slash is required
> and the $doc_root substitution should be replaced by ${doc_root} and
> there should be no slash following the }
> (just to have nice html without extra ////)
>

All reasonable. But, be aware that I have limited time, and no test
environment, for at least the next couple of weeks.