cvsweb and binary files

[paul gelderblom]

Daniel, others:

Daniel wrote:
> I also have a feature request.
> It would be nice with a search function in cvsweb.
> I did this as a separate cgi script but it would be more practical if it
was
> integrated into cvsweb.
> Just run find -name from the current directory and parse the output to
> get links back into cvsweb.

I built that!
Using File::find.
It  integrates with cvsweb (both ways: from cvsweb you can link to the
search functionality, allowing searches starting in a certain directory, and
from the search you get linked to the file in cvsweb)

It also does:
- content searches in RCS files or the latest revision (look for 'readme.*'
files containing 'Known Bug')
- regexes or just strings for the content match;
- filename selection using wildcards  patterns (*.cpp) or regexes
- case sensitivity or not for both match types
- selection based on edit status (watch on/off ;  edited files)
- selection based on file type (-kb -> binary )
and combinations of the above.

However:
- Due to the nature of it (blunt filesytem search) , it is a resource eater.
Maintainers on public repositories may not want it installed!
- I did not include taint checking (have only friendly users here). Thay may
even be complex, due to the user-specified regexes.

I do not think it should be integrated *in* CVSWEB since that is already
complex enough:
just integrate *with* cvswevb: crosslinks in both directions


Anyhow, enough blabla: should I post it?

Paul