PATCH: max-lisp-eval-depth exceeded on CVS Emacs

2005-09-21 Thread Jason Rumney
After a fresh install of Emacs and JDE on a new machine, I found that a 
patch I thought I'd submitted some time ago is not in either JDE 2.3.5 
or CVS.


The problem is caused by a change in behaviour of (expand-file-name .. 
dir) when dir is the root directory on a Windows drive. In previous 
versions, the result was D:/.., but in the current CVS code, 
(expand-file-name .. D:/) returns D:/, which is consistent with 
what happens on *nix.


The following patch adds support for this in jde-root-dir-p, avoiding 
some infinite loops:


*** c:/site-lisp/jde-2.3.5/lisp/jde.el~2004-12-16 22:29:36.0 
+
--- c:/site-lisp/jde-2.3.5/lisp/jde.el2005-09-21 17:29:28.89250 
+0100

***
*** 1595,1601 
   (or
(string= parent //) ; for paths like //host/d/prj/src
(string= parent ) ; for paths like \\host\d\prj\src
!(string= (substring parent -3) /..))) ; for paths like d:/prj/src
  ((member system-type '(cygwin32 cygwin))
   (or (string= (file-truename dir) (file-truename /))
   (string= parent //) ; for paths like //host/d/prj/src
--- 1595,1602 
   (or
(string= parent //) ; for paths like //host/d/prj/src
(string= parent ) ; for paths like \\host\d\prj\src
!(string= (substring parent -3) /..)
!(string= parent dir))) ; for paths like d:/prj/src
  ((member system-type '(cygwin32 cygwin))
   (or (string= (file-truename dir) (file-truename /))
   (string= parent //) ; for paths like //host/d/prj/src




Re: cygwin emacs path problem

2005-04-19 Thread Jason Rumney
Paul Kinnucan wrote:
Hi Felix,
As I understand it, there are two versions of cygwin emacs:
 (A) a version of Unix Emacs modified specifically to run in the 
 Cygwin environment and 

 (B) the standard Unix version of Emacs compiled, using cygwin
 gcc, to run in the Cygwin environment
JDEE supports the A version. It does not support the B
version.  The reason for this is that years ago, long
before the B-version existed, some JDEE users asked for A
version support and some of those users actually contributed
code necessary to support the A version. Meanwhile, until
now, there has been no demand for B-version support.
 

Really this is a cygwin problem. A cygwin version of Emacs needs to be 
modified to fit its environment, as the 'A' version was. I'm not sure 
what happened to those patches, but it is unreasonable of the Cygwin 
maintainers to expect the maintainers of every other package to make 
changes to accomodate Cygwin's lack of fitting in with its environment 
so that they themselves can use unmodified code targetted at GNU/Linux.




Re: Auto-format code to width = 80

2004-12-14 Thread Jason Rumney
Mark D. Hansen wrote:
Can anyone tell me how to configure emacs/jde so that my java code will
auto-format to have column width 80?
Thanks,
Mark
 

auto-fill-mode



Patch: File menu in recent CVS Emacs no longer called 'files

2004-11-10 Thread Jason Rumney

The File menu has recently been renamed from 'files to 'file in Emacs
CVS. This has revealed a dependancy in JDEE that can easily be fixed
by using menu-bar-file-menu (which has existed since at least 19.27)
which also simplifies the code a little:


*** jde.el-orig Fri Oct 29 00:43:02 2004
--- jde.el  Wed Nov 10 23:09:59 2004
***
*** 1247,1258 
(when (fboundp 'add-submenu)
  (add-submenu '(File) val Insert File...)))
(let* ((mb (assq 'menu-bar global-map))
-  (files (assq 'files mb))
   (menu (if (fboundp 'easy-menu-create-menu)
 (easy-menu-create-menu
  (car val) (cdr val
   (menu-name (car val)))
! (define-key-after (cdr (cdr files)) [jde-new]
(cons menu-name menu)
'open-file)

--- 1247,1257 
(when (fboundp 'add-submenu)
  (add-submenu '(File) val Insert File...)))
(let* ((mb (assq 'menu-bar global-map))
   (menu (if (fboundp 'easy-menu-create-menu)
 (easy-menu-create-menu
  (car val) (cdr val
   (menu-name (car val)))
! (define-key-after menu-bar-file-menu [jde-new]
(cons menu-name menu)
'open-file)


Re: Feedback needed on w3m support

2004-10-04 Thread Jason Rumney
Paul Kinnucan wrote:
I'd also appreciate it if someone who is using w3m on Unix would
take a look at the URL that the JDEE generates for local files and
tell me what's wrong with it and what URL would be acceptable
to w3m.
 

There was a thread about this last week. The problem is a double slash 
between the host part and filename part of the URL. A patch was posted, 
which I think will break things on Windows, since it replaced 
//localhost/ with //.  I think a better solution is to look at the 
filename, and if it starts with a slash (Unix), use //localhost or 
//, and if it doesn't (Windows), then use //localhost/ or ///.



Re: JDE 2.3.4beta3: strange overlining

2004-05-11 Thread Jason Rumney
Eric M. Ludlam [EMAIL PROTECTED] writes:

I'm getting the weird effect shown in the attached GIF using JDE 2.3.4beta3
and cedet1.0beta2 on Emacs 21.3. All non-public methods are 'overlined',
which is really not pleasant. I guess it's some kind of highlighting of
non-public methods that has gone wrong - does anybody know of a
configuration setting to modify this behaviour or is it possibly a bug?
   [ ... ]

 That is part of `semantic-show-tag-boundaries-mode'.  It actually
 overlines any method over a certain length (In characters.) 

No, I think the original poster is talking about the highlighting of
private and protected methods. For some reason the default
face for private members is a cyan overbar (but colors for overbars
don't seem to work, on Windows at least), and rather than just over
the function name, it is over every line of the function - that is
annoying, and what is also annoying is that if you customize it to
off and save, next time you start Emacs, the customization does not
take effect because it is executed before semantic-turn-on-everything
(or useful-things).

The doc string says that the default face is a background color,
incidently.


Re: Avoid JDEE to shadow standard libraries

2004-05-03 Thread Jason Rumney
David PONCE [EMAIL PROTECTED] writes:

 +   (condition-case nil
 +   ;; If the library if available, use it.
 +   (require feature)
 + (error
 +  ;; Try to use the one from the JDEE's distribution.
 +  (require feature (format jde-%s feature)

It might be cleaner to use:

(if (not (require feature nil t))
(require feature (format jde-%s feature)))



Re: JDE 2.3.3 craches emacs 21.3 (windows XP)

2004-02-19 Thread Jason Rumney
Gustavo Tenrreiro wrote:

Hi,

I just upgraded to emacs 21.3 for windows, and I find that at startup when emacs finds 
(require 'jde) on my .emacs file, it crashes. I get an abort dialog where I can 
either abort, retry, or ignore. In any case emacs goes down.
Is this a known problem? I ve tried on 2000 at home on a different machine, and the 
same happens.
Does anyone know what the story is here?
If you got the Windows binaries from ftp.gnu.org, then there is a 
problem with them. See recent posts to the help-emacs-windows mailing 
list archives at http://mail.gnu.org/ for links to working binaries.




Re: XEmacs and Eclipse

2003-12-20 Thread Jason Rumney
Shashank [EMAIL PROTECTED] writes:

 But recently looks like it has lost momentum for supporting new
 features for development like testing, J2EE ?

Since you are replying to me and Michael Coughlan, I can only assume
this is in response to my earlier reply to Michael. If so, you have
misunderstood my reply. All I said was that XML editing is not JDE's
responsibility, not that Emacs does not support it. There are many
people that use XML that are not interested in Java and vice-versa, so
it makes sense to keep the support for XML and Java seperate.




Re: IPlanet serverside debugging?

2003-04-03 Thread Jason Rumney
[EMAIL PROTECTED] writes:

 We can't control how it starts, but we still might be able to
 connect to the running process.
 
 You should be able to control how iAS starts jvm's e.g. see
 
 http://docs.sun.com/source/816-7149-10/dgdebug.html#25712

Unfortunately those docs are for Sun ONE app server 7, which is a
complete rewrite based on Tomcat, and bears little resemblance to iAS 6.

Last I checked, Sun had purged their webserver of references to old
iPlanet documentation, or at least made it hard to find.




Re: IPlanet serverside debugging?

2003-04-02 Thread Jason Rumney
Galen Boyer [EMAIL PROTECTED] writes:

 Is this doable within the jdee?

Which iPlanet? App Server and Web Server are completely different
products. I think iAS uses a standard JVM, in which case using
the remote debugging options would work. But I think iWS uses its own
JVM (from the Kiva days) and implements part of the Servlet API as
native code, so it might be more difficult.

If you are trying to debug something general in your servlet, running
it under Tomcat (or JBoss if you need more than just servlets/JSPs)
is probably easiest. If you are trying to debug something specific to
iPlanet, and can't run under a debugger, System.out.print() is your
friend.





Re: IPlanet serverside debugging?

2003-04-02 Thread Jason Rumney
Paul Kinnucan [EMAIL PROTECTED] writes:

 If you are trying to debug a program running in a vm embedded in
 another application, the container application must provide some means
 for you to tell it to start the embedded vm in debug server or client
 mode.

Often there is somewhere to specify command-line options to Java in
shell scripts/batch files that start the app server, or occasionally
in the config interface.




Re: Where is the make command in Cygwin

2002-05-09 Thread Jason Rumney

Riccardo Gianninoni [EMAIL PROTECTED] writes:

 Hi,
 I installed the Cygwin to compile this software:
 
   site
 jde
 speedbar
 semantic
 elib
 eieio
 
 but I can't find the make command.

Try to find a mingw32 port of make.  Cygwin programs do not
interoperate with other Windows tools very well, which is not good
for a program like make.


-- 
Jason Rumney




Re: Fix for a problem in saving project file (2)...

2002-02-25 Thread Jason Rumney

[EMAIL PROTECTED] (Ping Liang) writes:

 I have never needed to deal with Windows/Unix issues before, so I am
 not familiar with directory-sep-char.  I guess that there must be a
 better way to deal with this if it is to be deprecated. 

It is deprecated in GNU Emacs. I am not sure of its status in XEmacs
(it appears it is still needed there).  The reason for deprecating it
is clearly illustrated by this particular problem.

I think the correct fix is to only bind directory-sep-char around the
one line of code that requires it:

(defun jde-find-project-file (dir)
  Finds the next project file upwards in the directory tree
from DIR. Returns nil if it cannot find a project file in DIR
or an ascendant directory.
  (let ((file
 (let (directory-sep-char ?/) ;; Override NT/XEmacs setting
   (find jde-project-file-name
 (directory-files dir) :test 'string=
(if file
(expand-file-name file dir)
  (if (not (jde-root-dir-p dir))
  (jde-find-project-file (expand-file-name ../ dir))

-- 
Jason Rumney




Re: Needs help to get JDEE working...

2002-01-31 Thread Jason Rumney

Nandan Chandrakant Joshi [EMAIL PROTECTED] writes:

 Hallo,
 I have a problem getting JDE Menu in taskbar working. I have installed it as
 per the instructions and even .emacs file is configured accordingly. Well, here's 
some elementary info.:
 

I would suspect this section:

(defun my-java-mode-hook ()
(cond (window-system
   (require 'andersl-java-font-lock)
   (turn-on-font-lock)
  )
)
 )
 (add-hook 'java-mode-hook 'my-java-mode-hook)


Or maybe the lazy-lock line at the bottom;  jit-lock (the default) is
better supported in 21.1.

If commenting that out does not help, try commenting out sections
until you find what is causing it to fail.


-- 
Jason Rumney




Re: Whay is JDE/Xemacs startup so slow?

2002-01-28 Thread Jason Rumney

Paul Kinnucan [EMAIL PROTECTED] writes:

 Altmann, Michael writes:
   Do other people find that it takes 20 secs to load jde?  This seems
   excessively long.  I am running xemacs 21.1 on a solaris server box (I'm not
   sure its exact speed).  Without jde it takes just a second to start xemacs.
   There are many messages about loading regexp-opt on the status line.  Is
   there some sort of debug I should turn on to track what it is doing?
  
 The JDE is a big, complex application that requires about 30 lisp
 packages

It may still be worth debugging, since regexp-opt should only be loaded
once.  The total time to load JDE on my system is 4 seconds, although
if the solaris box has the directory where JDE lives mounted via NFS,
or is very old, 20 seconds might be reasonable.


-- 
Jason Rumney




Re: Extending JDE using Java (WAS: Gosling interview)

2002-01-27 Thread Jason Rumney

Daniel Hegyi [EMAIL PROTECTED] writes:

 I think this would be great! JDE is already a very powerful IDE, that
 is, it has a great deal of functionality. But I'm concerned with the
 UI aspect of it. Emacs-ers are used to doing M-x all the time,
 however, I think it could benefit much from some GUI components
 instead of having to memorize so many functions.

You sound like you are unaware of the menu-bar that has been a
feature of Emacs since the 18.x days.

I don't remember the last time I needed to do anything in JDE that
wasn't accessible from the menu-bar (even if I did find it quicker to
use the keyboard).

-- 
Jason Rumney




Re: Help Needed Debugging Tomcat Servelets

2002-01-26 Thread Jason Rumney

Heather Marie Buch [EMAIL PROTECTED] writes:

 I'm sorry, but could you please give me the syntax for setting it in
 startup.sh?
 
 I tried just writing the line 
 
 TOMCAT_OPTS=value

try adding the following line as well

export TOMCAT_OPTS

and make sure you are starting Tomcat with startup.sh, not
tomcat.sh start.

 Also, even when I set the TOMCAT_OPTS in tomcat.sh and start tomcat, and
 it echos that the opts
 are set to the debugging stuff, when I run netstat -a it doesn't show my
 tomcat socket number, so
 is tomcat really running on that socket?

I don't know. I would expect the socket to show up in netstat -a.
But the real test is if JDEbug can attach to the process.

 Is this what I would do?
 
 startup tomcat and apache with debugging opts
 go into emacs, open the java file
 start jdedebug
 attach process
 set breakpoint
 open servlet in browser

Yes.

 then what?

Then wait for your breakpoint to be hit.

 Or would I open the servlet in the browser in between attaching and
 setting the breakpoint?

That would work too, unless you are trying to debug the servlet's init
method, since that only runs the first time you connect.

 So, it looks like now I just need a way to make sure tomcat is starting
 on that socket, and the proper order to run the debugging.

Hopefully the above will help.

-- 
Jason Rumney





Re: Help Needed Debugging Tomcat Servelets

2002-01-26 Thread Jason Rumney

Heather Marie Buch [EMAIL PROTECTED] writes:

 BTW - At no point to I get the 
 
 step into or step over choices. Shouldn't they be available at some
 point after
 I attach the process?

They are available once Tomcat is stopped (ie a breakpoint is hit).

-- 
Jason Rumney