On Thursday 03 July 2003 07:52 pm, Greg Tyrelle wrote: [...] > I had the opportunity to introduce BioPHP in a 5 minute lightning > talk. The project generated a fair amount of curiosity from people at > BOSC, mainly because people are interested in using PHP to build > bioinformatics web applications, no great surprise. Oh, good, that means my position as "token wacko advocating non-web-based use of PHP as well" is likely to remain unchallenged by anyone else. :-) > [...]Keep in mind I > prepared these slides the night before, so apologies if there were any > mistakes etc. I don't know that it's necessarily MISTAKES, but I'm curious what you meant by "slow" and "lack of library support"? (Slow compared to compiled languages, or do you find PHP to be slower than, e.g., Perl?) By the latter, were you just referring to the fact that PECL is MUCH smaller than CPAN?... > Slightly more significant for the project as a whole was talking to > Chris Dagdigian of the open-bio.org foundation. I discussed with him > the possibility of BioPHP being supported by open-bio and the answer > was definitely yes. That means mailing lists, cvs repositories, DNS > and web hosting. He also offered biophp.open-bio.org if we want it ? > > Personally I think it sounds like a good offer, what do you all think ? I wouldn't mind "spreading" BioPHP project presence over several different sites (wider "net" means "capturing" more users and developers...) but I don't know that it's necessary to move things off of bioinformatics.org. (Bioinformatics.org did offer domain hosting of biophp.net, as I recall [actually they'd offered biophp.org, but that had gotten snapped up by the guys in Norway, who I haven't heard from in many months...]) Bioinformatics.org was the first organization to notice my little website [before Serge and I noticed each other's work] and step forward (unsolicited, no less) with offers of support for BioPHP. Maybe it'd be worth discussing how we might split up aspects of the whole project over several hosters if we wanted to (e.g. development/cvs on sourceforge, mailing lists/polls/experimental code [e.g. PHP5, Java-integration, etc.] on bioinformatics.org, release files and tutorials on open-bio.org, etc...). Don't know if it's really NECESSARY, but it'd give us wider exposure (and somehow the idea of encouraging as many interested organizations into contributing support is kind of appealing to me.) > And finally, that quote above is from Brad Chapman's BOSC blog (Brad > being one of the main biopython developers). It was just stream of > consciousness so I'd take it with a grain of salt :) Ah, okay, not what I'd guessed. I do get the impression that Python developers tend to be EXTREMELY loyal to the language. 'course, I imagine that if *I* had gone through most of the "popular" languages (which tend to be somewhat similar to one another) and found they didn't suit me as well as I'd like, and then hit the very different Python and found that it DID suit me well, I'd be exactly the same way... (Disclaimer - I actually built an "iterate through a FASTA file and return results of NCBI BLAST queries on each" sample application using BioPython as a way of trying out Python in general. I found Python just didn't suit me, though the application worked just fine. About the time I did that, you can see a post I made in the general BioPython mailing list in reply to someone mentioning the concept of BioPHP (which I THINK is the earliest reference on the 'net to the possibility of such a thing)...Naturally the response from the BioPython list didn't express much enthusiasm for the idea. :-) ("Just use Zope.") ) > In general the response to BioPHP was good, not "Why BioPHP ?" but > "Tell me more"... Well, thanks for introducing us there! Sounds like things went well, and I'm glad people seem interested. The timing's good, too, because I think we're approaching the point where a variety of real-world applications of the project will be feasible. I'm going to TRY to get an initial working draft of the export module finished sometime this weekend, and EFetch for sequences shortly thereafter. If there are no objections to my "BioPHP URLs" idea, I can even do 'extensions' of Sequence EFetch and NCBI-Blast that are import modules in the existing scheme... (Should there be a "NetIO" section for the core EUtils modules and such in the CVS tree?) Here's looking forward to next year, when BioPHP is wildly popular and has a whole "contingent" of users and developers at BOSC who travel in luxury due to vast sums of money donated by wealthy benefactors...hey, it could happen. :-) Sean