cvs cross reference tool

[Simon 'corecode' Schubert]

On Wednesday, 9. February 2005 04:04, Max Laier wrote:
> If you don't know it yet:
> http://fxr.watson.org/
>
> The search features from there are quite handy, didn't find anything like
> it in your tool yet.

yes, i know. i might link to rwatson's site, but i don't intend to do the same 
thing here: fxr indexes on identifiers, I index on files and revisions.

> Also, a quick check of the pf source (what else) didn't turn up the cool
> feature you were talking about (i.e. filling in the $OpenBSD versions)
> http://oly.corecode.ath.cx/~corecode/cgi-bin/crosscgi.py?cmd=detail&sourcei
>d=34780
> http://oly.corecode.ath.cx/~corecode/cgi-bin/crosscgi.py?cmd=detail&sourcei
>d=83740

yes, that's expected :/ openbsd and netbsd don't include the complete path in 
the tags, so my tool can only guess. and if there is more than one file in 
the corresponding repo with the same name, it will opt for manual fixing. 
there is already code in the cgi for that, though i'm not yet sure on how to 
cope with authorisation for changes.

> It's really fast, though!  How long does an index run take? On what
> hardware? How much space does it need for metadata?  Does it index
> incremental?

yes, i explicitly optimized it for speed. took some time :)
an index run for all repos takes less than one day i think, but i'd have to 
try again. the hardware is my desktop here in my room, a duron 1200 with 
384MB ram, and in fact it's at the moment updating the index (in parallel!), 
which is (after i cleaned out tagname references, whose tracking basically 
destroys all update/insert performance):
-rw-r--r--   1 corecode  wheel   672692224 Feb  9 04:14 cvscrosscheck.db

oh and, the changesets are created on fly, from aggregation to checkout.

i'm sure it could still be tuned, but i'm quite happy with the speed now. i 
had to write an own, fast rcs file parser, tho (and accompaning python 
module)


once i got my local changes into the repo again, i'll follow up with the 
source, if there is interest. oh and of course i'll post to the other 
projects' mailing lists, just wanted to make a first test here.

cheers
  simon

-- 
/"\
\ /
 \     ASCII Ribbon Campaign
/ \  Against HTML Mail and News
Attachment:
pgp00010.pgp
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pgp00010.pgp
Type: application/octet-stream
Size: 187 bytes
Desc: "Description: PGP signature"
URL: <http://lists.dragonflybsd.org/pipermail/kernel/attachments/20050208/f9dbfb91/attachment-0020.obj>