CEDET

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: Jake Colman <colman@pp...> - 2009-05-08 02:33 Henry, Thanks for those instructions. They worked fine. I was not entirely clear on the correct way to edit cogre-mode.el even after reading the .diff file. I tried to patch as follows: patch < cogre-mode.diff but it failed on all 3 hunks. What am I doing wrong? The build procedure looks ugly and it gives the appearance of not having worked. But it eventually completes without error and find . -name "*loaddefs* does seem to list all the files. Now I have to figure out how to make this work! ...Jake >>>>> "HST" == Henry S Thompson <ht@...> writes: HST> I regularly build CEDET from CVS using XEmacs (on Cygwin too, I'm a HST> glutton for punishment). HST> Here's what works for me at the moment >> cd [the root of your CEDET tree] >> cvs update >> find . -type f -name '*.elc' | xargs -n 100 rm >> cd semantic/symref >> cvs update -r1.5 symantic-symref-grep.el HST> Patch cogre/cogre-mode.el with the attached patch (you can do this HST> easily by hand if you don't understand how to run patch). >> cd ../cogre >> cvs update -d tests templates # these are new since I last updated >> cd .. >> make EMACS=xemacs all HST> The reversion of symantic-symref-grep is necessary because a breaking HST> change for XEmacs, namely a shift to using the 'grep' package, which HST> is internal to Emacs, I believe, and for which I could not quickly HST> find an XEmacs port . . . HST> The edit to cogre-mode is similarly necessary because of a recent HST> XEmacs-incompatible addition to this file. HST> Hope this helps, HST> ht HST> Index: cogre-mode.el HST> =================================================================== HST> RCS file: /cvsroot/cedet/cedet/cogre/cogre-mode.el,v HST> retrieving revision 1.25 HST> diff -b -w -C3 -r1.25 cogre-mode.el HST> *** cogre-mode.el 19 Apr 2009 00:33:06 -0000 1.25 HST> --- cogre-mode.el 7 May 2009 11:02:26 -0000 HST> *************** HST> *** 205,210 **** HST> --- 205,211 ---- HST> cogre-link-base-menu) HST> (defvar cogre-tool-bar-map HST> + (if (fboundp 'tool-bar-add-item) HST> (let ((tool-bar-map (make-sparse-keymap))) HST> (tool-bar-add-item "cogre-node" 'cogre-new-node 'node) HST> (tool-bar-add-item "cogre-class" 'cogre-new-node 'class) HST> *************** HST> *** 215,221 **** HST> (tool-bar-add-item "cogre-arrow" 'cogre-new-link 'arrow) HST> (tool-bar-add-item "cogre-isa" 'cogre-new-link 'inherit) HST> (tool-bar-add-item "cogre-hasa" 'cogre-new-link 'aggregate) HST> ! tool-bar-map) HST> "The tool-bar used for COGRE mode.") HST> --- 216,222 ---- HST> (tool-bar-add-item "cogre-arrow" 'cogre-new-link 'arrow) HST> (tool-bar-add-item "cogre-isa" 'cogre-new-link 'inherit) HST> (tool-bar-add-item "cogre-hasa" 'cogre-new-link 'aggregate) HST> ! tool-bar-map)) HST> "The tool-bar used for COGRE mode.") HST> *************** HST> *** 275,281 **** HST> (setq major-mode 'cogre-mode HST> mode-name "Cogre") HST> (use-local-map cogre-mode-map) HST> ! (set (make-local-variable 'tool-bar-map) cogre-tool-bar-map) HST> (setq truncate-lines t) HST> (setq indent-tabs-mode nil) HST> (buffer-disable-undo) HST> --- 276,282 ---- HST> (setq major-mode 'cogre-mode HST> mode-name "Cogre") HST> (use-local-map cogre-mode-map) HST> ! (if cogre-tool-bar-map (set (make-local-variable 'tool-bar-map) HST> cogre-tool-bar-map)) HST> (setq truncate-lines t) HST> (setq indent-tabs-mode nil) HST> (buffer-disable-undo) HST> -- HST> Henry S. Thompson, School of Informatics, University of Edinburgh HST> Half-time member of W3C Team HST> 10 Crichton Street, Edinburgh EH8 9AB, SCOTLAND -- (44) 131 650-4440 HST> Fax: (44) 131 651-1426, e-mail: ht@... HST> URL: http://www.ltg.ed.ac.uk/~ht/ HST> [mail really from me _always_ has this .sig -- mail without it is forged HST> spam] HST> ------------------------------------------------------------------------------ HST> The NEW KODAK i700 Series Scanners deliver under ANY circumstances! Your HST> production scanning environment may not be a perfect world - but thanks to HST> Kodak, there's a perfect scanner to get the job done! With the NEW KODAK i700 HST> Series Scanner you'll get full speed at 300 dpi even with all image HST> processing features HST> enabled. http://p.sf.net/sfu/kodak-com_______________________________________________ HST> Cedet-devel mailing list HST> Cedet-devel@... HST> https://lists.sourceforge.net/lists/listinfo/cedet-devel -- Jake Colman | Director, Software Development Principia Partners LLC 101 West Elm Street | Conshohocken | PA 19428 | +1 (610) 755 9770 t: +1 (610) 755 9786 | c: +1 (610) 348 2788 | f: +1 (201) 221 8929 e: colman@... | w: http://www.principiapartners.com Credit technology innovation awards winner 2008

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: Eric M. Ludlam <eric@si...> - 2009-05-08 03:01 Hi, I applied your cogre patch for compiling on XEmacs. I'm not sure about the symref issue. There were big issues trying to get those tests to pass on cygwin which is handled via that tool. Based on your description, I would guess something like this would help. *** semantic-symref-grep.el.~1.6.~ 2009-03-08 11:36:07.000000000 -0400 --- semantic-symref-grep.el 2009-05-07 22:14:44.000000000 -0400 *************** *** 29,35 **** ;; can be used in small projects to find symbol references. (require 'semantic-symref) ! (require 'grep) ;;; Code: --- 29,37 ---- ;; can be used in small projects to find symbol references. (require 'semantic-symref) ! (condition-case nil ! (require 'grep) ! (error nil)) ;;; Code: I'm always happy to apply XEmacs related patches to help things compile there, though I don't have a very recent XEmacs to test on. Eric >>> ht@... (Henry S. Thompson) seems to think that: >I regularly build CEDET from CVS using XEmacs (on Cygwin too, I'm a >glutton for punishment). > >Here's what works for me at the moment > >> cd [the root of your CEDET tree] >> cvs update >> find . -type f -name '*.elc' | xargs -n 100 rm >> cd semantic/symref >> cvs update -r1.5 symantic-symref-grep.el > >Patch cogre/cogre-mode.el with the attached patch (you can do this >easily by hand if you don't understand how to run patch). > >> cd ../cogre >> cvs update -d tests templates # these are new since I last updated >> cd .. >> make EMACS=xemacs all > >The reversion of symantic-symref-grep is necessary because a breaking >change for XEmacs, namely a shift to using the 'grep' package, which >is internal to Emacs, I believe, and for which I could not quickly >find an XEmacs port . . . > >The edit to cogre-mode is similarly necessary because of a recent >XEmacs-incompatible addition to this file. > >Hope this helps, > >ht > -- Eric Ludlam: eric@... Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: <klaus.berndl@ca...> - 2009-05-08 07:30 Hi guys, I suggest the following change for semantic--format-colorize-text: (defun semantic--format-colorize-text (text face-class) "Apply onto TEXT a color associated with FACE-CLASS. FACE-CLASS is a tag type found in `semantic-face-alist'. See this variable for details on adding new types." (if (featurep 'font-lock) (let ((face (cdr-safe (assoc face-class semantic-format-face-alist))) (newtext (concat text))) (put-text-property 0 (length text) 'face face newtext) newtext) text)) With this change at least the un-colorized text is returned if font-lock is not loaded - with the old implementation in this case nil was returned - and IMHO this is not what the user wants... With this change using all the semantic-format-functions work also when called with COLOR-arg not nil but font-lock not enabled/loaded - then they simply do no colorizing but at least they display something (with current impl they display nothing ;-) Thoughts? Thanks a lot! Klaus

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: Eric M. Ludlam <eric@si...> - 2009-05-08 11:13 I applied your patch. Thanks! Eric >>> Klaus seems to think that: >Hi guys, > >I suggest the following change for semantic--format-colorize-text: > >(defun semantic--format-colorize-text (text face-class) > "Apply onto TEXT a color associated with FACE-CLASS. >FACE-CLASS is a tag type found in `semantic-face-alist'. See this variable >for details on adding new types." > (if (featurep 'font-lock) > (let ((face (cdr-safe (assoc face-class semantic-format-face-alist))) > (newtext (concat text))) > (put-text-property 0 (length text) 'face face newtext) > newtext) > text)) > >With this change at least the un-colorized text is returned if font-lock is not loaded - with the old implementation in this case nil was returned - and IMHO this is not what the user wants... With this change using all the semantic-format-functions work also when called with COLOR-arg not nil but font-lock not enabled/loaded - then they simply do no colorizing but at least they display something (with current impl they display nothing ;-) > >Thoughts? > >Thanks a lot! >Klaus > -- Eric Ludlam: eric@... Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: Eric M. Ludlam <eric@si...> - 2009-05-08 13:21 >>> CVS with XEmacs seems to think that: >-----BEGIN PGP SIGNED MESSAGE----- >Hash: SHA1 > >Eric M. Ludlam writes: > [ ... ] >> >> (require 'semantic-symref) >> ! (condition-case nil >> ! (require 'grep) >> ! (error nil)) >> >> ;;; Code: > >OK, I tried that, and of course it now compiles. Not sure why this is >a good thing -- won't it just crash if anybody tries to _use_ it? In GNU Emacs, if the functions that are being used from grep.el don't exist, it falls back to an older technique (from the version of the file you were using.) Thus, if you can't load in grep.el at all, it will do the same thing, and everything ought to be ok. >> I'm always happy to apply XEmacs related patches to help things >> compile there, though I don't have a very recent XEmacs to test on. > >So I did make itest, which worked, and make utest, which failed as >follows: > > . . . > * Scheme tests... > symbol: define > symbol: fun1 > symbol: define > symbol: fun2 > * Scheme tests... > * ERROR: (error "Items forgotten: (\"fun1\" \"fun2\")") > >But this utest didn't seem like it would ever involve symref. . . The test suite exercises a lot of dependencies. This is likely just something else problematic that needs to be debugged. >Oh yes, you need the following in utest for XEmacs: > >(unless (boundp 'temporary-file-directory) > (defvar temporary-file-directory (temp-directory))) Thanks. I applied something similar. >I haven't dug into that failure -- suggestions? > >So I also tried M-x semantic-ia-utest, which failed _every_ test with > > .... Actual nil > >On setting debug-on-error to true, I got the following: > >Debugger entered--Lisp error: (error "Method clone called on nil") [ ... ] > clone(nil) > semantic-analyze-find-tag-sequence-default(("p") nil prefixtypes unfindable) > semantic-analyze-find-tag-sequence(("p") nil prefixtypes unfindable) [ ... ] > >Investigation revealed that the relevant buffer (testdoublens.cpp) had >no semanticdb-current-database, so semantic-calculate-scope was >returning a nil scope. I eventually forced one by hand, after which >the first two tests failed with > > ** Starting tests in testdoublens.cpp > Failed 1. Desired: ("pMumble" "publishStuff") Actual ("publishStuff") > Failed 2. Desired: ("pMumble" "publishStuff") Actual ("publishStuff") > >but then we lost with a nil scope again. . . > >Have I just jumped into the middle of something too hastily? How is >semanticdb mode supposed to be enabled in general? [ ... ] I changed the unit test suite recently to force semanticdb to ignore any old caches. That's if you run from the command line. If you run them from an interactive Emacs then it is configured normally. Anyway, this way the unit tests all run in a bootstrapping mode and tests that behavior. It seems likely that semanticdb is just not being initialized properly. This could happen if the semantic initialization on a per-buffer basis isn't working, which in turn might be related to how autoload files are created. I poke at the autoloads because only someone else is having trouble getting XEmacs to build them. Were your semanticdb issues related to an interactive XEmacs? If so, do you use semantic-load-enable-minimum-features in your .emacs ? Eric -- Eric Ludlam: eric@... Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: Eric M. Ludlam <eric@si...> - 2009-05-08 20:51 >>> Henry S. Thompson seems to think that: >OK, one more patch, and I can at least run the tests to completion. Thanks. I applied it as it seems like a good idea. >Still 3 failed tests, but I wonder if all the tests succeed for _you_? I run the utest suite before every submission or patch I mail out. I run the itest suite only if I change EDE or something deep in SRecode. I also try to write new tests whenever I make a new feature. Not everything is easily tested though, and I only got into the habit recently, so there is a lot of old stuff with no tests at all. I only do so for Emacs 23 though. I run the test suites many many times before putting a new .tar.gz out on all sorts of Emacsen and OSes at home and at work. Eric -- Eric Ludlam: eric@... Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net

Re: [CEDET-devel] Trouble Bu8ilding ceded from CVS with XEmacs From: Eric M. Ludlam <eric@si...> - 2009-05-14 02:10 >>> CVS with XEmacs seems to think that: >> The test suite exercises a lot of dependencies. This is likely just >> something else problematic that needs to be debugged. > >OK, I learned a lot about how the bovinator works, and uncovered the >problem (which as far as I can see means the Scheme tests shouldn't >ever have worked for anyone . . .): The scheme grammar doesn't >include numbers as expressions. Patch attached to fix this (and an >unrelated pblm with the definition of scheme symbols). Hi, Sorry for the long delay in replying. I've been busy. :) I've applied your patches. Thanks! Could you investigate if the semantic-lex-symbol-or-keyword analyzer would work in your case if you converted : into a symbol part? You would need to add code like this to the setup fcn for scheme. (setq semantic-lex-syntax-modifications '((?: "_") )) Chars like a or b are word syntax. _ is a syntax that is part of a symbol, but isn't part of a word. I gather from your patch that you just needed to add : to the list of _ like characters. If that works, I can undo parts of your patch. It would be very handy if you could augment cedet/semantic-utest.el for the scheme test to include a symbol using this syntax to exercise these symbols and number support you added. Thanks! Eric -- Eric Ludlam: eric@... Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net

1 message has been excluded from this view by a project administrator.