Beehive ate my previous attempt at sending this mail apparently.

-------- Forwarded Message --------
From: Tim Foster <[email protected]>
To: Danek Duvall <[email protected]>
Cc: [email protected]
Subject: Re: [pkg-discuss] Code review: make pkglint deterministic
Date: Fri, 07 Jan 2011 18:34:41 +1300

On Thu, 2011-01-06 at 15:09 -0800, Danek Duvall wrote:
> Tim Foster wrote:
> > http://cr.opensolaris.org/~timf/pkglint-deterministic-webrev

Thanks for the review.

> I don't know that it matters much here, but I would bet that sorting
> strings is faster than sorting objects, so something like
> 
>     [f.get_fmri() for f in sorted(fmris)]
> 
> is probably better written as
> 
>     sorted((f.get_fmri() for f in fmris))

Yes. I ran a few tests earlier, sorting large lists of FMRIs - it does
seem to be a bit better, 3 seconds faster against a list of 100,000
PkgFmri objects on my machine.  Hopefully pkglint never has to produce
an error message of that magnitude :-)  For more complex objects, it
might matter more of course.

It's a simple change though, so I included it.

> Also, in pkglint_manifest.py, in the first hunk, you treat
> undefined_variants and unknown_variants differently -- the first gets an
> explicit list and the latter doesn't.

Fixed that, thanks.

> Looks fine otherwise.

cool,

        cheers,
                        tim


_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to