On 2021/05/07 15:23, Chris Bennett wrote:
> On Fri, May 07, 2021 at 08:47:07PM +0100, Stuart Henderson wrote:
> > > I don't want to make a big mess on the mailing list, so I think I'll
> > > just submit these as WIP to the list as what the ports are called
> > > and put the files for each group of new ports elsewhere for anyone to
> > > look over first who wants to. Way too messy otherwise. This time I'm
> > > going to get properly organized first.
> >
> > Please send the tars on list rather than putting them elsewhere and
> > asking people to look at them.
>
> I was thinking more along the lines of just posting the Makefiles, at
> least at first, and if they are OK, then put up the tars.
tars, please, so we can actually build them.
if I have 5 minutes free to look at a ports submission, I'll be looking
for something where everything that I need to review it is in the mail.
> Is that OK or a bad idea? I'm just thinking that I'll get some simple
> stuff wrong in the Makefiles the first time and leave a mess on the
> list. I've never worked on a project this big, so just let me know
> anything that's the best way.
>
> A couple of the PGObject Makefiles were a bit complicated with the
> postgresql testing parts. Our postgres testing module even needed a diff
> for one of my Makefiles to work.
that's totally fine, the tree works together as a whole, sometimes other
parts (either of ports or sometimes even src) need a bit of work in
order to get a port working.
> I also have p5-PGObject-Util-DBAdmin, which I haven't submitted yet, that
> seems to need an interactive test that uses the postgres user's password
> and also has
> pre-test:
> mkdir ${WRKSRC}/t/var
>
> The testing part from the port isn't creating the var directory.
> portcheck complains about hardcoded paths for that.
The tools aren't always correct; portcheck whines about some things that
aren't a problem; sometimes update-plist gets things wrong. Best thing
to do in that case usually is to flag them in the mail.
> I'm good with whatever works best.
> Thanks,
> Chris
>
>
