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

Re: Perl QA Hackathon 2011 in Amsterdam

On 2011-04-19 11:04, Stefan Hornburg (Racke) wrote:
> Nice! Please make the input field larger for Perl modules with
> longer names.

size="60" now

On 2011-04-19 11:11, gregor herrmann wrote:
> Some quick thoughts:
> - For the link to the package I'd rather use
>   http://packages.qa.debian.org/$package than p.d.o. (p.qa.d.o is for
>   the source package and more developer-oriented)
>   Maybe the links to bugs and patches are not needed then, too? Or
>   maybe they are still helpful as shortcuts ...

link changed and patches+bugs kept for the moment. can be removed any time later

> - For the patches link you could maybe link directly to the
>   respective package version.


> - When searching for not-yet-packaged modules (like Debian::Apt::PM
>   :)) there's a typo: "use `reportbug` and fill-in RTP" -> ".. file
>   an RFP". For "join [the] Debian Perl Group" it's maybe more helpful
>   to link to http://wiki.debian.org/Teams/DebianPerlGroup or even
>   http://wiki.debian.org/Teams/DebianPerlGroup/Welcome

typos fixed, link is now http://wiki.debian.org/Teams/DebianPerlGroup/Welcome

> - I guess Gabor and the other CPAN guys will also need a flat file or
>   a web service, but he will speak up himself for sure :)

all the information is in the generated json files:


the name of the file is the first 2 characters of md5sum of module name:

perl -MDigest::MD5=md5_hex -le 'print substr(md5_hex("DBI"),0,2), ".json"'

On 2011-04-20 11:30, Jeremiah Foster wrote:
> Can it be made case insensitive? I think people will search for "moose" which
> produces no results and not "Moose" which produces results.

that is still possible with the md5 hashing that I use, but I was thinking that
it will be cool if someone creates a jquery cpan autocomplete plugin. so that
anyone who will have an input field with cpan module names in the future will
just load this plugin javascript and apply it on the input field so that there
are suggestions comming in.

On 2011-04-19 11:43, Dominique Dumont wrote:
> It can even find Perl modules in a more global distribution.

yes, the indexes are made from all .deb files found in the Debian mirror. it
took ~2 days to decompress and index them from scratch. then the incremental
indexes are fine. my next step will be to improve the scripts so that I don't
have to keep the full mirror as it takes a lot of disk space (92G => all+i386).
once indexed the .deb file is not needed. for now the trade of diskspace for my
time is good. :)

I hope to get to write the detailed description about how it works during the

Looks like the incremental runs destroyed the Distribution+Component records so
those were missing the days after Tuesday, now they are back again.


Reply to: