2007/07/31
IRC [10:27] *** Spark joined the chat.
IRC [10:28] <Spark> hi, i'm looking for a top-level description of the aolserver source code, what modules/files do what, and so on
IRC [10:28] <Spark> does such a thing exist?
IRC [10:38] <Dossy> Hmmm, not sure.
IRC [10:39] <Dossy> There may be such a doc, but it's likely outdated now. What purpose are you looking for this for?
IRC [10:41] *** dekoy joined the chat.
IRC [10:44] <dekoy> good morning
IRC [10:51] <Spark> Dossy: a year or so ago, someone used aolserver for a case study in lock inference, so they converted the source code to use atomic sections instead of manual locking
IRC [10:52] <Spark> i'm also trying to do this, so i can compare our lock inference techniques to there's, using the same code base
IRC [10:52] <Dossy> Interesting. I don't recall hearing about this--where was this/what was it for?
IRC [10:53] <Spark> http://berkeley.intel-research.net/dgay/pubs/06-popl-autolocker.pdf
IRC [10:54] <Spark> also http://www.cs.ucla.edu/~mje/publications/popl2007-talk.pdf
IRC [10:54] <Spark> oops that's the talk i think
IRC [10:54] <Spark> this is probably better http://www.cs.ucla.edu/~fischer/papers/popl07_locking.ps
IRC [10:55] <Dossy> interesting.
IRC [10:55] <Spark> i wrote to the authors, and they linked me to their modified source
IRC [10:55] <Spark> i think all the changes are in nsd/*
IRC [10:56] <Spark> but i'm particularly interested in tclvar.c, if you look at the table near the end of the second link, you can see their analysis took a long time
IRC [10:56] <Spark> page 6 of http://www.cs.ucla.edu/~fischer/papers/popl07_locking.ps
IRC [10:56] <Dossy> Let me look
IRC [10:58] <Dossy> I wonder why.
IRC [10:58] <Spark> yeah
IRC [10:59] <Dossy> I'm not exactly sure why AOLserver would be an interesting application to study here, as it's supposed to be free-threaded already.
IRC [11:00] <Dossy> Ah, I see... "We test two hypotheses about lock allocation:
IRC [11:00] <Dossy> first, many real programs use very simple locking disciplines
IRC [11:00] <Dossy> that should be automatically inferable; and second, our algorithm
IRC [11:00] <Dossy> performs efficiently in practice."
IRC [11:00] <Spark> this is some vague justification in section 6.3 (page 10) of http://berkeley.intel-research.net/dgay/pubs/06-popl-autolocker.pdf
IRC [11:01] <Spark> clearly the reason they used it in the 2007 paper was that it had already been annotated by different people for the 2006 paper :)
IRC [11:08] * Dossy nods.
IRC [11:08] <Dossy> It's always good to see folks building science on top of other's work, progressing it further.
IRC [11:09] <Dossy> It's also good to see that folks recognize AOLserver's academic value.
IRC [11:10] <Spark> it's quite a clean codebase, from what i've seen so far
IRC [11:14] <dekoy> hi guys just a quick question, any good reference/url on handling cross site scripting and sql injection in aolserver/tcl
IRC [11:14] <Spark> i was hoping to be able to understand at least bits of this code without learning tcl's C interface (i don't even know the tcl language) but now i think that was probably naive
IRC [11:15] <Spark> e.g. i'm not getting very far at all with "LockArray" in tclvar.c
IRC [11:16] <dekoy> ns_quotehtml
IRC [11:22] *** holycow joined the chat.
IRC [11:28] <dekoy> another quick question, what's the difference between ns_set get and ns_queryget, does ns_queryget still need ns_quotehtml to clean it ?
IRC [11:36] <Dossy> I need to run.
IRC [11:36] <Dossy> I'll be happy to answer your questions later--send me an email to dossy@panoptic.com
IRC [11:38] <dekoy> thanks
IRC [12:44] *** partymola parted the chat.
IRC [15:25] *** holycow parted the chat.
IRC [16:38] *** holycow joined the chat.
IRC [18:30] *** dekoy parted the chat.
IRC [19:33] *** partymola joined the chat.
IRC [20:06] *** partymola parted the chat.
IRC [20:07] *** holycow parted the chat.
IRC [23:38] *** holycow joined the chat.