Re: Patch name changes

[Alistair Crooks <agc%pkgsrc.org@localhost>]

On Sun, Jun 13, 2010 at 05:07:13PM +0200, Thomas Klausner wrote:
> On Sun, Jun 13, 2010 at 06:39:29AM +0200, Alistair Crooks wrote:
> > On Thu, Jun 10, 2010 at 08:58:36PM +0200, Thomas Klausner wrote:
> > > On Thu, Jun 10, 2010 at 08:01:57PM +0200, Alistair Crooks wrote:
> > > > Interesting syntax - is there any prior art for this kind of naming 
> > > > scheme?
> > > 
> > > Yes, both FreeBSD and OpenBSD use it.
> > 
> > OK, thanks.  Not sure why you clipped the rest of my mail, or what the
> > objection to a standardised scheme is...
> 
> I'll answer the other question first.
> 
> > Also, can you remind me again the driver for the change - i.e. what's
> > the problem statement?  It's not obvious from your original mail.
> 
> I wrote in my original email:
> > This makes it easier to see from the file system, what files are
> > patched; it also helps packagers of other porting systems to more
> > easily identify e.g. from cvsweb at which files to look. 

To be insanely accurate, that's not a problem statement.  It's a goal,
perhaps laudable.  Leaving the situation as is will not cause anything
to fail.  Implementing it may help, and so some CBA is needed first -
cost versus benefit - to see whether it's a worthwhile thing.
 
> So my aim was better readability of patch file names for packagers
> (from NetBSD or other systems). I don't think that strvis, even
> http-style, is preferable, since "patch-path%2fto%2ffile" is much
> harder to read than "patch-path_to_file" and we don't need the full
> functionality of strvis since we just need to quote '/'.

I actually agree with your visibility argument; but I also feel we've
become lax in our quoting styles, and Reply-To forwarding that means
that sometimes answers get missed.

I also have a problem with dictats - the "everyone must do packaging
THIS way" form of edicts, which snuff out creativity, and lead to
corporate mundane sheep-like behavior. Please allow people to innovate.

Regards,
Alistair