| |
| |
| |
|
Page: 1 2 3 4 5 6 7
Comments:
<0> mega1: I'm not sure I understand the question. the stuff that's currently #ifdef x86/x86_64 only?
<1> yes
<1> in other words, what is the reason that ppc/gencgc doesn't use it?
<0> not the fast_bzero or the sniff_code_object stuff
<0> we don't need to do the code fixups on ppc, right?
<0> you're probably talkinb about the scav_vector stuff and friends
<2> hmm, am i misunderstanding something, or is (format t "~,4F" 1.0e-5) => "0.00001" in sbcl wrong
<0> talking, even
<2> every other lisp except cmu i tried gives 0.0000 which i'd expect
<1> slyrus: yes, sorry
<3> beach: I take it you're back? :)
<0> mega1: I don't remember why we can (or if we can) get away with the abbreviated non-x86-oid scav_vector I see there
<0> rtoy might remember what he did for CMUCL here
<1> slyrus: the x86 version is just for additional performance for eq based hash tables.
<1> but is there a reason for not using the same code?
<0> not that I can think of, but I haven't really thought about it
<0> mega1: you want me to try it?
<1> slyrus: thanks, not yet. I'll post an untested patch to sbcl-devel instead a bit later since I'm changing cheneygc too.
<4> hi timoore (:
<0> ok. let me know. I'm happy to spin some builds.
<5> jamesjb: I think 0.0000 is the right answer.
<0> is richard newman related to WHN?
<6> slyrus: doesn't seem likely.
<7> ... Ugh. Damned windows documentation. Figuring out how to draw underlined text took far longer than it should. :-/
<0> ah, I'd imagine not. I forgot Richard Newman was holygoat.
<8> tfb: if you remind me tomorrow I can do some measurements on ppc startup times
<9> Xophe: thanks
<9> I'm getting about 0.4s round-trip for a top-level fn which does nothing now
<9> which I can live with but faster would be better
<9> (I dunno why I had 3s before)
<10> nyef: ping
<10> nyef: is your Windows UI code in terms of SBCL's FFI?
<7> Yes, it is. Why?
<7> You think CFFI is triggering that intermittent GC problem?
<10> nyef: just curious. No, I'm wondering if you'd be inclined to give Graphic-Forms a try (I'm planning a release at the end of this week).
<7> I'm a little leery of some of your dependencies.
<7> (Looked through your source control system a little while ago...)
<10> I've reduce the dependencies this time around. (no more Cells, for one thing).
<7> Heh.
<7> Guess I'll have to take a look, then.
<7> I've also been reading up on CLIM recently.
<10> :-) ok...you'll see the announcement when it goes out, any comments you might have are very welcome.
<7> So I don't know what I'm going to end up doing yet.
<10> also, ImageMagick is optional now (I have a plugin system for graphics libraries)
<7> Actually, I'm toying with the idea of writing my own CLIM implementation.
<7> Or at least part of one.
<4> hah
<10> nyef: oh, cool. For Windows, you mean?
<7> antifuchs: Given some of the other projects on my list, what's one more obnoxiously complex and underspecified one?
<4> something that's clim with an eye on the interactions between application frames and command loops and so on would be neat (:
<7> Yeah, for windows.
<7> Also possibly for X, but for windows first.
<4> nyef: haha. sure you don't want to do my pathnames homework? (:
<7> Fairly sure.
Return to
#lisp
or
Go to some related
logs:
#math
truetype fonts on Debian
#python
dli hdd driver
#css
#math
#kde
#kde
#nvidia
htaccess php4-cgi php_value
|
|