10:31 a.m.
I've just checked large parts of the new GC infrastructure into CVS.
This has been tested quite thoroughly under Unix/Xt, but not at all
under Gtk and Windows.
Therefore, I would tremendously appreciate if some kind souls running
Windows or Gtk would update their CVS sandboxes, configure with
--use-kkcc, fire up the resulting XEmacs and suffer through a few GCs,
say, by starting Dired on a large directory or something.
Please report (either to xemacs-beta or to me) ESPECIALLY if things go
smoothly.
--
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla
11:39 a.m.
sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) writes:
Therefore, I would tremendously appreciate if some kind souls
running
Windows or Gtk would update their CVS sandboxes, configure with
--use-kkcc, fire up the resulting XEmacs and suffer through a few GCs,
say, by starting Dired on a large directory or something.
I would love to do so, but 21.5 doesn't compile on windows since several
weeks.
--
One Ring to rule them all, One Ring to find them,
One Ring to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.
12:29 p.m.
Frank Schmitt <ich(a)Frank-Schmitt.net> writes:
sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr.
Preprocessor]) writes:
> Therefore, I would tremendously appreciate if some kind souls running
> Windows or Gtk would update their CVS sandboxes, configure with
> --use-kkcc, fire up the resulting XEmacs and suffer through a few GCs,
> say, by starting Dired on a large directory or something.
I would love to do so, but 21.5 doesn't compile on windows since several
weeks.
Likewise for GTK, patch below needed. Last time I tried using a GTK
compiled XEmacs it looked so horrible (many display bugs, distorted
minibuffer, toolbar etc) so it is rather unusable for more than a 1
minute session.
Index: objects-gtk.c
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/src/objects-gtk.c,v
retrieving revision 1.9
diff -u -p -u -w -r1.9 objects-gtk.c
--- objects-gtk.c 2002/06/20 21:18:38 1.9
+++ objects-gtk.c 2002/07/29 11:26:49
＠＠ -425,7 +425,7 ＠＠ gtk_font_spec_matches_charset (struct de
/* find a font spec that matches font spec FONT and also matches
(the registry of) CHARSET. */
-static Lisp_Object gtk_find_charset_font (Lisp_Object device, Lisp_Object font,
Lisp_Object charset);
+static Lisp_Object gtk_find_charset_font (Lisp_Object device, Lisp_Object font,
Lisp_Object charset, int stage);
#endif /* MULE */
11:59 p.m.
sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) writes:
Therefore, I would tremendously appreciate if some kind souls
running
Windows or Gtk would update their CVS sandboxes, configure with
--use-kkcc, fire up the resulting XEmacs and suffer through a few GCs,
say, by starting Dired on a large directory or something.
Please report (either to xemacs-beta or to me) ESPECIALLY if things go
smoothly.
As options in win32 are set in nt/config.inc, I could only guess about
how to set this option. I included a line
USE_KKCC=1
this seems to have been the right way, so I propose to add the following
lines to nt/config.inc.samp
#uncomment this if you want to use the new Garbage Collector
#USE_KKCC=1
Well, I couldn't really find a difference in behavior until I set
gc-cons-threshold to 1000 (it was set to 3000000 before), then I had the
impression, that the XEmacs compiled without the KKCC-option is slightly
(perhaps 10-20%) faster than the one compiled with USE_KKCC=1.
I'll just do a posting to build-report including my config.
--
One Ring to rule them all, One Ring to find them,
One Ring to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.
8:45 a.m.
>>>> "Frank" == Frank Schmitt
<ich(a)Frank-Schmitt.net> writes:
Frank> As options in win32 are set in nt/config.inc, I could only guess about
Frank> how to set this option. I included a line
Frank> USE_KKCC=1
Frank> this seems to have been the right way, so I propose to add the following
Frank> lines to nt/config.inc.samp
Frank> #uncomment this if you want to use the new Garbage Collector
Frank> #USE_KKCC=1
I'll do that if nobody objects. Many thanks for the note!
Frank> Well, I couldn't really find a difference in behavior until I set
Frank> gc-cons-threshold to 1000 (it was set to 3000000 before), then I had the
Frank> impression, that the XEmacs compiled without the KKCC-option is slightly
Frank> (perhaps 10-20%) faster than the one compiled with USE_KKCC=1.
Yes, that seems about right: The new one allocates somewhat more storage.
--
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla
8151
days inactive
8152
days old
5 comments
3 participants
participants (3)
-
Frank Schmitt -
Michael Sperber -
Simon Josefsson