Hello, 

just a gentle reminder on this one. Is it OK?

Regards

Enzo

On Mon, Jan 05, 2026 at 02:52:32PM +0000, Enzo Nicosia wrote:
> Hello Stuart, 
> 
> thanks a lot for your comments and sorry for the late reply.
> 
> Thanks for the simpler tgz. I only made a change in DESCR to note that
> qrqscore would need p5-libwww in order to run. Amended tgz attached, and
> responses to your points are inline below.
> 
> Is this OK?
> 
> Enzo
> 
> On Tue, Dec 30, 2025 at 10:46:48AM +0000, Stuart Henderson wrote:
> > I would drop the flavour. My understanding is that the OSS emulation in
> > libossaudio shouldn't be used for new things - the preferred option is
> > patching to use sndio - sio_open(3) etc - bit otherwise I would use the
> > existing pulseaudio support.
> 
> I agree on removing the FLAVORS, if the OSS emulation is considered
> legacy. The pulseaudio version was the original one I built. Then I
> realised that the OSS version compiled and ran seamlessly with minimal
> changes, and no added deps. That's why I tried the version with two
> FLAVORS. I will see if it is easy to get a native sndio version in the
> future, but I think it is safer to include only the pulseaudio option
> for now. 
> 
> > 
> > Various comments,
> > 
> > - option-specific patches are a pain to deal with, if the flavour was
> > necessary everything you've done in the patch variants could be handled
> > by a smaller patch + MAKE_ENV/MAKE_FLAGS instead
> > 
> 
> OK, point taken. I have now read again about MAKE_ENV/MAKE_FLAGS, and I
> see what you mean. That makes total sense. 
> 
> > - generally don't patch for strlcpy snprintf etc in ports, if done at all
> > that should go upstream (but also should check return values etc).
> > patching to use PATH_MAX+300 for a string holding a path makes no sense.
> > i'd just drop that patch
> > 
> 
> OK, I will then make a proper patch and upstream it to the author
> instead of including it in the port.
> 
> > - qrqscore needs p5-libwww to run. not sure if it's worth adding a dep.
> > a note in DESCR might be appropriate but considering the code quality
> > (besides the string handling, there's unsafe /tmp use, calls to system
> > for things that woukd be better done from C, etc) I'm not sure I'd
> > want it going near data fetched from the net
> >
> 
> I have added a note in DESCR about that. Well, makes a single request
> to a page owned by the qrq author. I would probably leave it like it is.
> 
> > - no need to patch for 'make uninstall', it's never called from ports
> >
> 
> point taken. Thanks.
> 
> > possible simpler tar attached
> 
> 
> 
> -- 



-- 

Reply via email to