All the Perl that's Practical to Extract and Report

search.cpan.org code is not open

acme on 2004-06-29T10:04:15

One of the things that annoys me (there are many, this is just one) is that the code that runs search.cpan.org is not available. By available, I mean access to the version control system it uses is available and that releases are available on CPAN. If it was, then people would be able to hack on it and it would be so much better. For example, its search would not suck.

This annoyance has been brought to you with blue M&Ms and some furry Tigger ears.

Yes, I know about kobesearch.cpan.org aka CPAN::Search::Lite, but it has no tests. Also note that I generally work on fixing things that annoy me.

Very nice ears

ambs on 2004-06-29T11:02:33

Yeah, the tiger ears are very nice :) and, of course, orange!

feel the same way

TeeJay on 2004-06-29T11:24:37

The code should be available, then we could see how to improve it.

One of the biggest annoyances is that the stemming of words is over-enthusiastic and that function names and the like aren't indexed.

Writing a good search engine in perl is pretty damned easy, especially when you have access to all the data.

If the code was in CPAN along with some testdata, then any of us could improve or replace it with something that works.

not to mention more sites could run it

hfb on 2004-06-29T11:40:13

After years late nights babysitting the original box and sometime worrying about the current one, it has been a very desireable goal to get something people could install on their own boxes and reduce the load. People like to say that search is cpan, but its not even close with 1 search box vs. 250 mirrors.

Re:not to mention more sites could run it

drhyde on 2004-06-29T15:25:13

search may not be cpan, but i bet it's how a fair proportion of the users use cpan.

Re:not to mention more sites could run it

hfb on 2004-06-29T16:47:34

If that were true, then the number of people using CPAN is a much smaller number than people like to think. And, besides, you totally missed the point.

Re:not to mention more sites could run it

bart on 2004-06-29T19:14:03

I've been wondering about this for a long time... Once you decide to download a package based on a search through search.cpan.org, does it forward you to one of the mirrors — preferable a close one?

If not, I'm having a hard time seeing the benefit of having the mirrors...

Re:not to mention more sites could run it

hfb on 2004-06-29T19:29:39

It does if you choose a mirror from the mirror menu...but the point is if there were more of the search boxes it could be done locally, i.e. behind corporate firewalls and elsewhere since it keeps a mirror resident on the box. There is still plenty of benefit from having mirrors around the world since cpan.pm and other tools use them...not to mention the other seach engines.

Actor

ambs on 2004-06-29T16:05:05

Leon, we are missing the main actor for our hack story :)

Have you asked?

samtregar on 2004-06-29T19:31:50

When I wanted to add a diff tool to search.cpan.org I just wrote to Graham and asked him if I could do it. He agreed and after a small amount of administrative wrangling I got onto the dev system and got to work. I found the code to be quite good overall and I had no trouble finding the appropriate place to begin.

Perhaps you can do the same?

-sam

... did you ask?

ask on 2004-07-01T21:56:36

ranting about it on use.perl is probably not nearly as efficient as asking Graham would be.

    - ask

Re:... did you ask?

acme on 2004-07-02T08:28:09

I've asked him before. I ask him every year or so. People complain about this all the time. Getting the source by asking isn't nearly good enough. See gnat's rant while you're at it.

I have decided that sometimes I'm just too nice. Instead, I'm getting things done and complaining to people. Let's see if this works.