[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [britney/master] Add a new hint file: satbritney


Am Mittwoch, den 28.03.2012, 23:24 +0200 schrieb Julien Cristau:
> On Wed, Mar 28, 2012 at 22:16:03 +0200, Joachim Breitner wrote:
> > Am Mittwoch, den 28.03.2012, 19:11 +0200 schrieb Julien Cristau:
> > > On Fri, Mar 23, 2012 at 15:33:22 +0000, Philipp Kern wrote:
> > > > SAT britney will generate additional easy hints akin to the autohinter.
> > > > They're currently generated on ries and committed into a git repository
> > > > which will be pulled regularly onto franck.
> > > > 
> > > So as far as I can tell this seems to generate entirely bogus hints.
> > > Do we really need more noise in update_output?
> > 
> > entirely bogus is a bit strong, as far as I know the main difference in
> > transition semantics between britney2 and SAT-Britney is that
> > SAT-Britney considers smooth transitions in all sections, not just
> > selected ones, hence the current hints related to java.
> > 
> The hint I'm seeing in the log right now is:
> Trying easy from satbritney: babl/0.1.6-2 libisfreetype-java/5.3.20100629-3 libisnativec-java/5.3.20100629+fix-1
> which unless I'm missing something is complete bullshit, babl has
> nothing to do with those other packages.

please, calling the result of someone else’s work complete bullshit is
not really encouraging contributions.

But if what you meant to saying is something like: „I’m surprised to see
that there is one hint involving both babl and the other package, would
you care to explain“, then I’ll happily do that. I’ll also move it to
the list so that everyone interested can read the explanation:

The hint file http://release.debian.org/britney/hints/satbritney lists
first the complete transition as figured out by SAT-Brintey; hence it
lists both babl and the java packages.

Then SAT-Britney tries to break it down to minimal nontrivial
transitions and generates hints from these, hence the next three hints.
Here, it detects that the hint only mentions one package and is likely
not accepted by britney and comments it own; this was requested by mehdi
during DebConf 11.

One may ask why the full hint is listed at all. The reason is that
SAT-Britney will only generate 5 minimal hints for now. This is a
safeguard as generating minimal hints is relatively slow and if there
were a situation that would allow for many minimal hints I don’t want
the runtime to explode.
Also, because one-package-hints are commented out, I would not get
feedback from britney why these hints fail. In the current setup, I can
see from britney messages after the full hint where SAT-Britney goes

Once I manage to get SAT-Britney in line with britney’s semantics (wrt
to smooth transitions and renamed packages and what else comes up), then
it should not be needed to generate smaller hints at all, as the large
hint will always work. But I’m not there yet.


Joachim "nomeata" Breitner
Debian Developer
  nomeata@debian.org | ICQ# 74513189 | GPG-Keyid: 4743206C
  JID: nomeata@joachim-breitner.de | http://people.debian.org/~nomeata

Attachment: signature.asc
Description: This is a digitally signed message part

Reply to: