[CHANGE PROPOSAL] Moving WWW from pkg-descr to Makefile
bapt at FreeBSD.org
Sat Oct 6 08:40:15 UTC 2012
On Sat, Oct 06, 2012 at 10:04:09AM +0200, Baptiste Daroussin wrote:
> On Fri, Oct 05, 2012 at 07:07:49PM -0700, Doug Barton wrote:
> > On 10/05/2012 07:05 AM, Baptiste Daroussin wrote:
> > > Hi,
> > >
> > > I want to propose and make the actual move of the WWW information from the
> > > pkg-descr to the Makefile itself via a WWW variable.
> > >
> > > doing this will have multiple benefits:
> > > * consistency all metadata bug this one are in Makefile
> > ... except for the distinfo data.
> > > * speedup make describe avoiding using grep to get the informations (make
> > > describe itself does not need speed but make index heavily use it and this
> > > will definitly benefit from speed up)
> > Agreed, but there are other ways to speed up 'make describe,' and tools
> > like ports-mgmt/p5-FreeBSD-Portindex are a much better solution where
> > "create an up to date INDEX quickly" is a real need.
> This sounds like: we have a problem, instead of fixing it let's workaround it
> > > * Third party tool will be able to probe the information more easily.
> > Portmaster doesn't parse WWW for anything, but I have suggested to
> > others in the past that 'while read' in sh is a faster method of finding
> > the WWW than grep. In fact, 'make describe' uses this method now, not
> > grep. I vaguely recall that this was done at my suggestion, but I'm too
> > lazy to trace the commit path to confirm it. :) 'make www-site' (which
> > already exists in case there is a 3rd party tool that needs it) uses a
> > different method utilizing awk. I haven't tested which one is faster.
> > I'm also concerned about the proliferation of things being jammed into
> > the ports Makefile (and coincidentally, bpm). I've noticed that just
> > about every operation that portmaster does using 'make -V' has gotten
> > noticeably slower over the last 6-8 months, and the trend seems to be
> > getting worse instead of better.
> make www-site already exists and will be made faster, and nothing else will be
> added, but things will be removed... I don't see the proliferation here but
> rather a cleanup.
> Concerning the bpm, I have removed more old things and useless tests from it
> that I have added! and if people were actively working on switching there ports
> to the new option framework then the fallback code could be removed and this
> will speed up lots of operations.
> > > Do anyone have any concern about this?
> > For all these reasons, and for the already-stated reason that it
> > severely lowers the value of 'cat pkg-descr' (which I do quite often,
> > and I'm sure other users do as well) I think this is a bad idea.
> > Doug
Here is a diff of the final impact on bpm (once WWW: is gone)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20121006/fc1b1778/attachment.pgp
More information about the freebsd-ports