On Fri, Jan 27, 2012 at 8:25 AM, Ketil Malde <ke...@malde.org> wrote: >> Using conduit is just great! > > From your example, it sure looks less noisy than most iteratee code I've > seen.
The biggest different that you'll see using conduits is that it's much easier to write library code. It's very very tricky to write an Enumeratee, but writing Conduits is dead simple. > I haven't gotten into iteratees beyond reading about it, mostly > since I have old libraries doing lazy parsing (with all the strictness > issues already ironed out), and that works well enough. But I'm > currently reviving my old BLAST XML parser, so perhaps I'll look at > fitting it to conduits? While I don't know exactly what you need for the BLAST XML parser, I definitely recommend you to take a look at xml-conduit [1], the streaming interface in particular [2]. As far as I can imagine, writing a 'Conduit Event m BlastXML' won't be difficult and will be pretty fast. BTW, Michael is trying to solve the biggest source of overhead that affects my biostockholm library, so take that 12 MiB/s as a lower bound. =) [1] http://hackage.haskell.org/package/xml-conduit [2] http://hackage.haskell.org/packages/archive/xml-conduit/0.5.1.2/doc/html/Text-XML-Stream-Parse.html > Anyway, although I'm not too familiar with the Stockholm stuff, it > sounds great! Feel free to update the wiki as well. It's on my TODO list already ;-). Cheers! =D -- Felipe. _______________________________________________ Biohaskell mailing list Biohaskell@biohaskell.org http://malde.org/cgi-bin/mailman/listinfo/biohaskell
