April 1999 - XEmacs-Beta - XEmacs Mailing List Archives

Re: Properties and readonly objects 25 years, 7 months

Hrvoje Niksic

Jan Vroonhof <vroonhof(a)math.ethz.ch> writes: > x-remap-keysyms-using-function-key-map which is uninterned. The > constants vector is the only reference. However this vector is in > pure space and thus is not swept on GC. Therefore there is no link > to the symbol and thus its gets GC. If the vector is in pure space, then all its symbol contents should be interned in Vpure_uninterned_symbol_table until the end of time, and GCPRO'ed through that. If that doesn't happen, it's a bug in the whatever code produces the pure compiled function. As far as I can tell from scanning the code, it's correct: DEFUN ("make-byte-code", Fmake_byte_code, 4, MANY, 0, [...] if (purify_flag) { [...] /* Statistics are kept differently for the constants */ if (!purified (f->constants)) { f->constants = Fpurecopy (f->constants); } This Fpurecopy() should intern the symbols to Vpure_uninterned_symbol_table. > I am not sure this what Martin actually fixed. You mean Kyle? > However, this is incomplete. What happens if I unintern it later (as > x-win-init-sun does)? If you unintern it later, it doesn't matter, because you still have a living reference in Vpure_uninterned_symbol_table, which should keep the GC sweeper from putting its greasy hands on the symbol.

1 participants
0 comments

Re: Properties and readonly objects 25 years, 7 months

Michael Sperber

>>>>> "Hrvoje" == Hrvoje Niksic <hniksic(a)srce.hr> writes: Hrvoje> sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) writes: >> >>>>> "Hrvoje" == Hrvoje Niksic <hniksic(a)srce.hr> writes: >> Hrvoje> BTW, why do we want Lisp-level readonly objects in the first Hrvoje> place? >> >> The most obvious is that, with the right dumper and OS support, you >> can put them in shareable/read-only/EPROM memory. Hrvoje> Isn't this the point of C-level readonly objects? If I understand Hrvoje> Olivier correctly, the "lisp-level readonly objects" would still be Hrvoje> modifiable from C, which would make them unsuitable for Hrvoje> shareable/read-only memory. Hrvoje> Or am I misunderstanding Olivier? Hmm, my understanding of Lisp-level readonly object was "readonly objects created from Lisp." But maybe I'm wrong. >> Another is that you can optimize GC with this kind of knowledge. It >> seems that, since reachability is affected by mutable plists, you >> want the plists of immutable objects to be immutable as well. Hrvoje> Hmm. The old "pure" stuff behaved this way for strings. It took the Hrvoje> easy way out for symbols, simply refusing to create pure symbols. Can't strings also have plists? -- Cheers =8-} Chipsy Friede, Völkerverständigung und überhaupt blabla

1 participants
0 comments

XEmacs 21.0-b66 "20 minutes to Nikko" failed to compile! 25 years, 7 months

Dan Glastonbury

Does any one have any suggestions on how to fix this? The same problem occurs when I try to compile XEmacs 21.2. I'm using egcs-1.1.2. [Snip!] hashtable storage: 12200 streams used: 2 streams on free list: 2 stream storage: 198 Testing for Lisp shadows ... Invalid read syntax: "Invalid integer constant in reader", "2", 10xemacs exiting.ggmake[2]: Leaving directory `/efs/zeeman/xemacs-21.0.66/src' ggmake[1]: Leaving directory `/efs/zeeman/xemacs-21.0.66/src' Building finder database ... Invalid read syntax: "Invalid integer constant in reader", "2", 10xemacs exiting.ggmake: *** [lisp/finder-inf.el] Error 255 Exit 2 XEmacs Configure Information ---------------------------- uname -a: IRIX gauss 6.2 03131015 IP22 ./configure '--site-includes=/usr/local/include' '--site-libraries=/usr/local/lib32' XEmacs 21.0-b66 "20 minutes to Nikko" configured for `mips-sgi-irix6.2'. Where should the build process find the source code? /efs/zeeman/xemacs-21.0.66 What installation prefix should install use? /usr/local What operating system and machine description files should XEmacs use? `s/irix6-0.h' and `m/iris4d.h' What compiler should XEmacs be built with? gcc -g -O3 -Wall -Wno-switch Should XEmacs use the GNU version of malloc? yes Should XEmacs use the relocating allocator for buffers? yes What window system should XEmacs use? x11 Where do we find X Windows header files? Where do we find X Windows libraries? Additional header files: /usr/local/include Additional libraries: /usr/local/lib32 Runtime library search path: /usr/local/lib32:/maptek/people/dan/carmen/shlib/SGI_IRIX.6.2.g++:/maptek/carmen/shlib/SGI_IRIX.6.2.g++:/maptek/vulcan_3.3_for_carmen_n32/lib:/mnt/root_gauss/local/qt-1.41/lib:/usr/freeware/lib:/usr/lib32/mips3:/usr/lib32 Compiling in support for XAUTH. Compiling in support for XPM images. Compiling in support for PNG image handling. Compiling in support for (builtin) GIF image handling. Compiling in support for JPEG image handling. Compiling in support for TIFF image handling. Compiling in native sound support. Compiling in support for DBM. Compiling in support for proper session-management. Using Lucid menubars. Using Lucid scrollbars. Using Motif dialog boxes. Compiling in DLL support. movemail will use "dot-locking" for locking mail spool files. Compiling in extra code for debugging. Compiling in code for checking XEmacs memory usage. WARNING: --------------------------------------------------------- WARNING: Compiling in support for runtime error checking. WARNING: XEmacs will run noticeably more slowly as a result. WARNING: Error checking is on by default for XEmacs beta releases. WARNING: --------------------------------------------------------- thanks Dan -- Dan Glastonbury <Dan.Glastonbury at maptek.com.au> "Two things are infinite: the universe and human stupidity; and I'm not sure about the the universe." -- Albert Einstein

1 participants
0 comments

Machine donations... 25 years, 7 months

William M. Perry

Is xemacs.org set up to take charitable donations at all? After the free-software-frenzy in japan, I really started annoying our CFO and got him to let me donate some older machines to the project of my choice. If XEmacs cannot use them, then I will probably just send them on to the FSF. -Bill P.

1 participants
0 comments

question about case in key maps 25 years, 7 months

John H Palmieri

XEmacs 21.2 "Demeter" [Lucid] (sparc-sun-solaris2.5) of Mon Mar 15 1999 on ahoward I posted this question to comp.emacs.xemacs, but I thought this mailing list might be a good place for it, also. I get this behavior with XEmacs 21.2b13, and I think also with some version of XEmacs 20.0. As described below, I know how to fix this in GNU Emacs, but not in XEmacs. Is this difference in the two emacsen a bug or feature? I'm having a problem defining a keymap in XEmacs; the quick version is: I want the key "a" to be bound to something, and I want "A" to be undefined. When I try to carry this out, though, hitting "A" just does whatever "a" is bound to. Details: I initialize the keymap with (defvar my-map nil) ;; initialize the keymap (define-prefix-command 'my-map) ;; initialize the keymap (define-key my-map "a" 'self-insert-command) ;; bind "a" I check to see how "A" is bound: (lookup-key my-map "A") > nil Then I test it: I bind backquote to my-map and see what happens. (global-set-key "`" 'my-map) Now hitting `a inserts a. Hitting `A inserts a. Hitting `b gives the message ` b is not defined. How can I get `A to act like `b? I've tried (define-key my-map "A" 'undefined) and (suppress-keymap my-map) Each of these does what I want with GNU Emacs, but not with XEmacs. -- John H. Palmieri palmieri.2(a)nd.edu ..I must be a VETERINARIAN..

1 participants
0 comments

Re: Properties and readonly objects 25 years, 7 months

Jan Vroonhof

Hrvoje Niksic <hniksic(a)srce.hr> writes: > This has been fixed long ago. Not good enough. I am pretty sure the x-win-init-sun crash reported for 21.0.67 is due to this. As I wrote before > But purified aggregate objects like lists and vectors can > contain uninterned symbols. If there are no other non-pure > references to the symbol, then the symbol is not protected > from garbage collection because the collector does not mark > the contents of purified objects. This is a good description of the problem > So to protect the symbols, > an impure reference has to be kept for each uninterned symbol > that is referenced by a pure object. All such symbols are > stored in the hash table pointed to by > Vpure_uninterned_symbol_table, which is itself > staticpro'd. */ And this is a good solution > if (NILP (XSYMBOL (obj)->obarray)) > Fputhash (obj, Qnil, Vpure_uninterned_symbol_table); > return obj; > } However, this is incomplete. What happens if I unintern it later (as x-win-init-sun does)? This code needs to set some flag such that unintern can add it to Vpure_uninterned_symbol_table later. Jan

1 participants
0 comments

Re: Properties and readonly objects 25 years, 7 months

Hrvoje Niksic

Jan Vroonhof <vroonhof(a)math.ethz.ch> writes: > Hrvoje Niksic <hniksic(a)srce.hr> writes: > > > > > Hmm. The old "pure" stuff behaved this way for strings. It took the > > easy way out for symbols, simply refusing to create pure symbols. > > > > And it breaks big time when you have reference to symbols in pure > objects. This has been fixed long ago. See Fpurecopy(): /* Order of subsequent tests determined via profiling. */ else if (SYMBOLP (obj)) { /* Symbols can't be made pure (and thus read-only), because assigning to their function, value or plist slots would produced a SEGV in the dumped XEmacs. So we previously would just return the symbol unchanged. But purified aggregate objects like lists and vectors can contain uninterned symbols. If there are no other non-pure references to the symbol, then the symbol is not protected from garbage collection because the collector does not mark the contents of purified objects. So to protect the symbols, an impure reference has to be kept for each uninterned symbol that is referenced by a pure object. All such symbols are stored in the hash table pointed to by Vpure_uninterned_symbol_table, which is itself staticpro'd. */ if (NILP (XSYMBOL (obj)->obarray)) Fputhash (obj, Qnil, Vpure_uninterned_symbol_table); return obj; }

1 participants
0 comments

Re: Properties and readonly objects 25 years, 7 months

Jan Vroonhof

Hrvoje Niksic <hniksic(a)srce.hr> writes: > > Hmm. The old "pure" stuff behaved this way for strings. It took the > easy way out for symbols, simply refusing to create pure symbols. > And it breaks big time when you have reference to symbols in pure objects. See the 'x-win-init-sun' crash messages for 21.0 on this list. Jan

1 participants
0 comments

Re: Majordomo Slurpage 25 years, 7 months

Colin Rafferty

So what is the URL of the Majordomo SlurPage? Will this page have random slurs, or just legitimate ones? -- Colin

2 participants
1 comments

Re: Properties and readonly objects 25 years, 7 months

Hrvoje Niksic

sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) writes: > >>>>> "Hrvoje" == Hrvoje Niksic <hniksic(a)srce.hr> writes: > > Hrvoje> BTW, why do we want Lisp-level readonly objects in the first > Hrvoje> place? > > The most obvious is that, with the right dumper and OS support, you > can put them in shareable/read-only/EPROM memory. Isn't this the point of C-level readonly objects? If I understand Olivier correctly, the "lisp-level readonly objects" would still be modifiable from C, which would make them unsuitable for shareable/read-only memory. Or am I misunderstanding Olivier? > Another is that you can optimize GC with this kind of knowledge. It > seems that, since reachability is affected by mutable plists, you > want the plists of immutable objects to be immutable as well. Hmm. The old "pure" stuff behaved this way for strings. It took the easy way out for symbols, simply refusing to create pure symbols.

1 participants
0 comments

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

XEmacs-Beta April 1999