2014-12-29T00:00:36Z oleo__ joined #sbcl
2014-12-29T00:00:36Z oleo is now known as Guest56975
2014-12-29T00:02:19Z oleo__ quit (Read error: Connection reset by peer)
2014-12-29T00:02:52Z Guest56975 quit (Ping timeout: 250 seconds)
2014-12-29T00:03:15Z oleo__ joined #sbcl
2014-12-29T00:03:33Z oleo__ is now known as oleo
2014-12-29T00:26:34Z eudoxia joined #sbcl
2014-12-29T00:37:52Z milosn quit (Ping timeout: 264 seconds)
2014-12-29T00:41:27Z attila_lendvai quit (Quit: Leaving.)
2014-12-29T00:43:01Z milosn joined #sbcl
2014-12-29T00:58:52Z stassats is doing (with-ir1-namespace (let ((*source-paths* (make-hash-table)) (*undefined-warnings*) (*lexenv* (make-lexenv :default (make-null-lexenv)))) (ir1-toplevel '(if x y) nil nil)))
2014-12-29T00:59:04Z stassats: with different forms and trying to tell heads from tails
2014-12-29T00:59:11Z stassats: so far not so successful
2014-12-29T00:59:25Z nyef: Maybe they're double-headed?
2014-12-29T01:06:30Z stassats: more like quadruple
2014-12-29T01:07:12Z karswell quit (Read error: Connection reset by peer)
2014-12-29T01:07:35Z karswell joined #sbcl
2014-12-29T01:12:07Z stassats: it doesn't seem human readable, perhaps i should just start writing s-exp conversion code
2014-12-29T01:16:19Z nyef: Wasn't there some CLIM-based IR1 display tool?
2014-12-29T01:16:40Z stassats: i only heard about it, never seen
2014-12-29T01:18:59Z nyef: I found screenshots, and I think maybe source code at one point.
2014-12-29T01:19:07Z nyef: ... Within the past few months, IIRC.
2014-12-29T01:19:14Z nyef: But I don't remember offhand WHERE.
2014-12-29T01:19:22Z stassats: and that wouldn't really be helpful
2014-12-29T01:19:30Z stassats: "oh, see, someone else understood IR1"
2014-12-29T01:19:51Z pacon quit (Read error: Connection reset by peer)
2014-12-29T01:20:08Z |3b|: https://github.com/onixie/ir1-viewer ?
2014-12-29T01:22:06Z nyef: That might have been the code, but not the right author...
2014-12-29T01:24:13Z |3b|: probably http://www.lichteblau.com/blubba/irspect-2007-03-19.tgz then
2014-12-29T01:24:23Z nyef: Yeah, that's a lot more likely.
2014-12-29T01:24:24Z |3b|: since that one has pictures too
2014-12-29T01:24:30Z stassats: i don't irspect it to work with the current sbcl
2014-12-29T01:34:38Z pacon joined #sbcl
2014-12-29T01:43:37Z scymtym quit (Ping timeout: 258 seconds)
2014-12-29T01:47:22Z stassats: i understand all the different nodes, i don't get the control flow
2014-12-29T01:53:28Z nyef: What's the issue with control flow?
2014-12-29T01:53:50Z stassats: what is it?
2014-12-29T01:55:20Z stassats: nodes are linked together, and clbocks are linked together
2014-12-29T01:55:46Z nyef: By way of CTRANs?
2014-12-29T01:55:59Z stassats: including ctrans
2014-12-29T01:56:30Z stassats: but nodes have NEXT/PREV as well
2014-12-29T01:57:02Z stassats: so it's node->ctran->node? what's the point of the ctran?
2014-12-29T01:57:38Z nyef: It's an indirection layer, for starters.
2014-12-29T01:57:43Z stassats: and when i have a cblock, how do i get the first node? it seems to only have the last one, do i just walk backwards?
2014-12-29T01:58:12Z nyef: A CTRAN doesn't have a notion of a "previous" node, so it can be used more than once.
2014-12-29T01:58:21Z nyef: Err... Not "used", but transferred to.
2014-12-29T01:58:46Z nyef: Your CBLOCK has a BLOCK-START, which is the initial CTRAN.
2014-12-29T01:59:11Z nyef: (See also CTRAN-STARTS-BLOCK in IR1UTIL.)
2014-12-29T01:59:21Z stassats: ok, i missed start
2014-12-29T02:00:24Z stassats: and i found what ir1-toplevel returns to be not really enlightening, so i redid it with returning the component
2014-12-29T02:00:37Z stassats: so that i can see that start block and cetera.
2014-12-29T02:01:29Z nyef: You're messing with one of the parts of the compiler that I understand the least.
2014-12-29T02:02:17Z stassats: there's also this DFO
2014-12-29T02:03:28Z nyef: Yeah. DFO and earlier is where my understanding really breaks down. That, and the while constraint / type-propagation thing.
2014-12-29T02:11:58Z eudoxia quit (Quit: Leaving)
2014-12-29T02:21:11Z stassats: i guess i see now
2014-12-29T02:21:50Z stassats: now to sleep on it
2014-12-29T02:28:49Z nyef: Sleep well.
2014-12-29T02:29:13Z stassats: the horros of CTRANS between REMs
2014-12-29T02:29:32Z echo-area joined #sbcl
2014-12-29T03:38:29Z christoph_debian quit (Ping timeout: 252 seconds)
2014-12-29T03:51:23Z christoph_debian joined #sbcl
2014-12-29T04:04:05Z pkhuong: ctrans can't be transferred to multiple times
2014-12-29T04:04:24Z pkhuong: they are mostly useful to factor out splicing logic.
2014-12-29T04:05:06Z nyef: ... They can't?
2014-12-29T04:06:29Z nyef: Ah, I see. They are largely internal to cblocks, so you transfer to the cblock rather than to the start-ctran.
2014-12-29T04:06:35Z pkhuong: ctrans capture evaluation order, blocks capture control flow and nodes capture dataflow.[C[D[D[D[D[D[D[D
2014-12-29T04:11:01Z pkhuong: lvars + nodes for data flow, ctrans + nodes for evaluation order within a basic block.
2014-12-29T04:13:55Z nyef: Right... And the unfortunate bit with ctrans is that they basically prevent reordering things within a basic block.
2014-12-29T04:15:42Z pkhuong: no, that's their purpose
2014-12-29T04:17:40Z nyef: Okay, it's quarter past eleven, I have a head full of CLIM bits, and I'm supposed to be working tomorrow. Paging the basics of the compiler design back into my head is probably not a good idea right now.
2014-12-29T04:18:40Z krzysz00 quit (Quit: leaving)
2014-12-29T04:19:02Z joshe: nonsense
2014-12-29T04:19:12Z joshe: there is never a bad time to read about compilers
2014-12-29T04:19:58Z nyef: "You may say this, but you're wrong."
2014-12-29T04:20:35Z joshe: haha
2014-12-29T04:29:45Z psy joined #sbcl
2014-12-29T04:52:27Z krzysz00 joined #sbcl
2014-12-29T05:02:16Z sean_1 joined #sbcl
2014-12-29T05:07:15Z LiamH quit (Quit: Leaving.)
2014-12-29T05:07:35Z daviid joined #sbcl
2014-12-29T05:39:56Z scymtym joined #sbcl
2014-12-29T06:04:12Z nyef quit (Quit: G'night all)
2014-12-29T06:04:47Z pranavrc joined #sbcl
2014-12-29T06:05:20Z psy quit (Ping timeout: 244 seconds)
2014-12-29T06:25:20Z sean_1 left #sbcl
2014-12-29T06:32:39Z psy joined #sbcl
2014-12-29T07:06:54Z loke: Hello
2014-12-29T07:40:19Z daviid quit (Ping timeout: 272 seconds)
2014-12-29T07:42:40Z Bicyclidine quit (Ping timeout: 244 seconds)
2014-12-29T07:57:21Z redline6561 quit (Ping timeout: 252 seconds)