15:28:26 ccl-logbot [~ccl-logbo@setf.clozure.com] has joined #ccl 15:28:26 15:28:26 -!- names: ccl-logbot Fare ogamita rme segv- gleag alms_ patrickwonders @ChanServ Blkt peccu2 mdc PuffTheMagic asedeno fe[nl]ix xristos strawman0f brucem |3b| gz- jdz krrrcks pjb gbyers wws clop ivan`` 15:51:09 -!- wws [~billstcla@p-74-209-25-247.dsl1.rtr.chat.fpma.frpt.net] has quit [Quit: Linkinus - http://linkinus.com] 15:57:22 billstclair [~billstcla@unaffiliated/billstclair] has joined #ccl 16:11:00 rme: herep 16:11:10 rme: it looks like on ccl trunk, asdf is not precompiled anymore, but loaded from source. 16:11:36 that kind of sucks. 16:12:48 NB: I built ccl with (ccl:rebuild-ccl :full t) 16:13:28 Fare: Is there an actual practical difference? 16:17:28 loading time much reduced if precompiled. 16:19:41 Well, with CCL's compiler speed... 16:20:02 on my machine, .24 seconds vs 1.64 seconds. 16:21:00 a factor 6.8 16:21:45 and 80MB of allocation less. 16:22:54 Are you sure it's a trunk problem? 16:23:11 I see the same thing in the 1.9 release, for what it's worth. 16:23:25 I have an otherwise clean source tree 16:23:40 of trunk at r15893 16:24:04 do you have a fasl for asdf in tools/ ? 16:26:04 Nope. 16:26:17 Well, yes, now I do. :) 16:26:23 I did it by hand, though. 16:28:45 I mean, if require is supposed to handle things automatically, I'd expect it to handle the file compilation. 16:32:31 it doesn't 16:32:55 and in general, can't -- e.g. in a system-installed CCL where you don't have write permission to tools/ 16:33:33 CCL used to do it right. Somehow, there was bitrot. rme probably knows where to fix it. 16:34:30 What was the way of doing it right? 16:34:44 Fare: ASDF isn't compiled as part of rebuild-ccl. I made it quit doing that some time ago. I think it had to do with compile-time side-effects that left packages/symbols in the resulting image. 16:35:18 oh. Can you do it in a fork / subprocess ? 16:35:25 Fare: I guess that in system-installed CCLs, the installer ought to do it? 16:35:35 Or the install script... 16:36:15 -!- ogamita [~t@77.104.4.54] has quit [Ping timeout: 256 seconds] 16:36:19 (OK, you can't fork because of the way you use threads) 16:36:40 yes -- and I was thinking that rebuild-ccl is the perfect install script for that. 16:37:25 rme: how is ASDF worse than any contrib about creating packages, anyway? 16:37:38 You might want all your contribs to be compiled in a subprocess. 16:42:20 -!- Fare [fare@nat/google/x-zwhsuqfqiiwvreeo] has quit [Ping timeout: 260 seconds] 17:17:57 Vivitron [~Vivitron@c-50-172-44-193.hsd1.il.comcast.net] has joined #ccl 17:32:32 Fare [fare@nat/google/x-kccqqlwpamjwxwku] has joined #ccl 17:32:59 Fare: it deletes packages 17:34:20 fe[nl]ix, ASDF ? Oh, yes, in the worst case scenario, it does. But only packages that use a previous obsolete and incompatible of ASDF. 17:35:43 which in the case of CCL, means ASDF 2.26 or older. 17:37:18 (I don't remember why CCL can't, like SBCL, survive upgrade from older versions of ASDF without this drastic package blast — I believe it's related to issues with SETF FOO where FOO's package is renamed, and it might have been fixed in the latest CCL) 17:39:14 you're focusing too much on upgrading 17:39:38 an asdf.lisp that's included into a CL implementation doesn't need that 17:40:51 -!- segv- [~mb@95-91-242-160-dynip.superkabel.de] has quit [Ping timeout: 256 seconds] 17:47:44 segv- [~mb@95-91-242-160-dynip.superkabel.de] has joined #ccl 17:48:18 -!- segv- [~mb@95-91-242-160-dynip.superkabel.de] has quit [Remote host closed the connection] 17:49:20 then the code path won't be exercised 17:49:21 segv- [~mb@95-91-242-160-dynip.superkabel.de] has joined #ccl 17:49:31 and isn't to be scared by 17:51:42 also, the latest ASDF is more deterministic in its compile-time side-effects: nowadays, EVERYTHING is in an eval-when that gets evaluated at compile-time as well as load-time. 17:53:25 (that solves "interesting" problems that existed in previous versions of ASDF, that could only safely compiled after it was loaded already, leading to non-trivial construction rules that ASDF itself could follow, but that made it slightly harder to build from e.g. the ECL or CCL build scripts.) 18:51:03 -!- patrickwonders [~patrickwo@user-38q42ns.cable.mindspring.com] has quit [Quit: patrickwonders] 18:51:52 patrickwonders [~patrickwo@user-38q42ns.cable.mindspring.com] has joined #ccl 19:19:17 ipmonger [~IPmonger@pool-72-94-39-57.phlapa.fios.verizon.net] has joined #ccl 20:30:00 -!- Fare [fare@nat/google/x-kccqqlwpamjwxwku] has quit [Ping timeout: 245 seconds] 21:20:46 Fare [~fare@cpe-69-203-115-132.nyc.res.rr.com] has joined #ccl 22:08:02 -!- gleag [~gleag@71.175.broadband2.iol.cz] has quit [Ping timeout: 264 seconds] 22:52:14 -!- billstclair [~billstcla@unaffiliated/billstclair] has quit [Read error: Connection reset by peer] 23:02:18 -!- rme [~rme@50.43.147.243] has quit [Quit: rme] 23:26:37 -!- segv- [~mb@95-91-242-160-dynip.superkabel.de] has quit [Remote host closed the connection] 23:49:19 rme [~rme@50.43.147.243] has joined #ccl 23:59:06 -!- Fare [~fare@cpe-69-203-115-132.nyc.res.rr.com] has quit [Ping timeout: 240 seconds]