hi all, here is a wacky idea for a 'talk' for this tuesday. anyone here use or know of ack? ack is a pure perl replacement for the grep utility but with more features and such. i got involved on the side when i was asked to look at making it faster. i easily found that its central loop (line in, match and process, line out) was very convoluted, clunky and slow. i have come up with a general new design for this inner core and i know it will be much faster and easier to maintain than the current code.
ack's home page is http://petdance.com/ack/ so my proposal is a hackathon on this code. with a bunch of heads and hands we can take my general redesign and likely implement it right during the meeting. it isn't that much code, maybe 2-3 pages total that we would write. and it can be split into several chunks to be worked on in parallel. we could sync up to the ack source tree or work on our own branch or whatever. we would need someone (or a team) to manage all source control and access. someone could be the tester, another could be the main editor for merging code, transcribing my ramblings or whatever. another box could show the current code for all to look at. the rest can help with ideas, coding, heckling and lurking. someone could report results on irc #boston.pm or send out emails with the latest source. someone can act as the PHB (i won't volunteer for that). so this is for an existing and popular perl project. the rewrite is doable in the time frame (< 3 hours). i have the general design (several minor variations need to be hammered down) but it is open to change. it would be something to crow about at oscon and at other perl monger groups. :) this would be a different sort of meeting with most people working on some aspect of the project. we should do as much premeeting work as possible but i won't give out jobs unless this idea is accepted for the meeting. i guarantee you will learn some interesting perl techniques as i have several in my design. nothing outragious but not stuff you see everyday. one tip is to rip out code from Read::FileBackwards to become a forward buffered read for ack. that will be much faster than the way it does input now. another is a shift register that has references to the output lines. any thoughts? any volunteers? uri -- Uri Guttman ------ [EMAIL PROTECTED] -------- http://www.sysarch.com -- ----- Perl Code Review , Architecture, Development, Training, Support ------ --------- Free Perl Training --- http://perlhunter.com/college.html --------- --------- Gourmet Hot Cocoa Mix ---- http://bestfriendscocoa.com --------- _______________________________________________ Boston-pm mailing list [email protected] http://mail.pm.org/mailman/listinfo/boston-pm
