Incrementally building ports INDEX

[Ion-Mihai Tetcu]

On Tue, 24 May 2005 09:56:30 +0100
Matthew Seaman <m.seaman at infracaninophile.co.uk> wrote:

> On Fri, May 20, 2005 at 11:48:24PM +0300, Ion-Mihai Tetcu wrote:
>  
> > Kris, do you think you could run it on the cluster where the "official"
> > INDEX is generated for a few times and diff the resulting index with the
> > canonical one and send the diff (if any) to Matthew ? It won't place any
> > burden on the machine (on my home desktop, it runs 6 times / day after
> > each cvsup and it takes under one minute to generate the updated index).
> > Once we know it works you could disable it and re-enable it only when
> > testing bsd.*.mk changes in case one of them affects it (and then Mathew
> > would have enough time to fix the problem before the changes go in the
> > tree).
> 
> There will be some differences -- but nothing significant, I hope.
> When testing I found that the sort order generated by the BDB hash
> tree is subtly different to the sort order generated by sort(1).  Also
> canonical 'make index' collapses multiple spaces to single within
> package comment strings, which p5-FreeBSD-Portindex doesn't.

Yes, we've discussed these two, but I don't see what impact could they
have.

> > This tool is very nice to use, especially on less-powerful machines and
> > IMO deserves mentioning in the docs like the porttools.
> 
> Thank you very much for the vote of confidence, and for the comments
> and bug reports you've provided, which have been very helpful to me.

Well, I'm using it. And perl is somewhere down on my list of languages I
like / use so they where bug reports, not patches :-/


-- 
IOnut
Unregistered ;) FreeBSD "user"