"Recent" changes to compilation-mode?

"Recent" changes to compilation-mode?

[David Abrahams]

Sometime in the past year or so compilation-mode started acting very
strangely for me with a CVS-built emacs.  First, if I use next-error
while compilation is still going on, I will often, after a moment or
two, see the point jump unpredictably in the file it takes me to.

Second, next-error used to cause the error in question to scroll to
the top of the *compilation* buffer's window, which made it very easy
to see what was going on.  Now emacs just moves that buffer's point
without scrolling.  Since the active point ends up in the source
file's buffer where the error is, all I see change in the
*compilation* buffer is that an unfilled rectangle with super-thin
borders moves.  It's extremely hard to see; if the window isn't going
to scroll, IMO something else needs to be done (e.g. with
highlighting) to emphasize the error in question.

Finally, many of my carefully-constructed error regexps seem no longer
to be recognized.  I had been doing:

  (require 'compile)

  (setq compilation-error-regexp-alist
        (append
         '(...regexps...)
         compilation-error-regexp-alist))

   (if (fboundp 'compilation-file-regexp-alist) ; in case of XEmacs,
                                                ; which doesn't support
                                                ; 'compilation-file-regexp-alist
    (setq compilation-file-regexp-alist
      (append
       '(... regexps ...)
           compilation-file-regexp-alist)))

I note that now there appears to be a two-layer scheme with
compilation-error-regexp-alist-alist that renders my old settings
invalid.  I can adjust what I'm doing, but discovering what to do was
a struggle.  Is there a way to provide a better transition experience
for people who upgrade?

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

"Recent" changes to compilation-mode?

[Nick Roberts]

 > Second, next-error used to cause the error in question to scroll to
 > the top of the *compilation* buffer's window, which made it very easy
 > to see what was going on.  Now emacs just moves that buffer's point
 > without scrolling.  Since the active point ends up in the source
 > file's buffer where the error is, all I see change in the
 > *compilation* buffer is that an unfilled rectangle with super-thin
 > borders moves.  It's extremely hard to see; if the window isn't going
 > to scroll, IMO something else needs to be done (e.g. with
 > highlighting) to emphasize the error in question.


When there is a fringe present, the default was changed to not scroll
because the arrow points to the current error and errors immediately
above the current one remain visible.

When there is no fringe, the current error scrolls to the top of the
*compilation* buffer's window.

So you should either see an arrow or scrolling.  If you don't, you need
to give your configuration.  M-x report-emacs-bug, also on the menu bar
plus the value of fringe-mode and/or window-fringes is a good start.


Nick

P.S  To always scroll to the top: (setq compilation-context-lines 0)

Re: "Recent" changes to compilation-mode?

[David Abrahams]

Nick Roberts <nickrob@snap.net.nz> writes:

>  > Second, next-error used to cause the error in question to scroll to
>  > the top of the *compilation* buffer's window, which made it very easy
>  > to see what was going on.  Now emacs just moves that buffer's point
>  > without scrolling.  Since the active point ends up in the source
>  > file's buffer where the error is, all I see change in the
>  > *compilation* buffer is that an unfilled rectangle with super-thin
>  > borders moves.  It's extremely hard to see; if the window isn't going
>  > to scroll, IMO something else needs to be done (e.g. with
>  > highlighting) to emphasize the error in question.
>
>
> When there is a fringe present, the default was changed to not scroll
> because the arrow points to the current error and errors immediately
> above the current one remain visible.
>
> When there is no fringe, the current error scrolls to the top of the
> *compilation* buffer's window.
>
> So you should either see an arrow or scrolling.  If you don't, you need
> to give your configuration.  M-x report-emacs-bug, also on the menu bar
> plus the value of fringe-mode and/or window-fringes is a good start.

Yeah, there's a tiny arrow there.  The problem is that the diagnostics
typically span several lines and are often quite dense, and the arrow
isn't quite enough to help me distinguish the boundaries of the
diagnostic being indicated by the arrow.  Would it be possible to add
some kind of highlighting to the current error, similar to what I get
when hovering the mouse over it?

> P.S  To always scroll to the top: (setq compilation-context-lines 0)

Ah, that will be a big help, thanks.  Not quite as good as
highlighting, but not bad.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

Re: "Recent" changes to compilation-mode?

[Juri Linkov]

> Yeah, there's a tiny arrow there.  The problem is that the diagnostics
> typically span several lines and are often quite dense, and the arrow
> isn't quite enough to help me distinguish the boundaries of the
> diagnostic being indicated by the arrow.  Would it be possible to add
> some kind of highlighting to the current error, similar to what I get
> when hovering the mouse over it?

Recently I proposed a patch that does exactly what you want, i.e.
highlights the whole current error message with a special user option.
You can get the patch here:

http://lists.gnu.org/archive/html/emacs-devel/2005-05/msg00561.html

This patch is too good for the coming release.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: "Recent" changes to compilation-mode?

[David Abrahams]

[-- Attachment #1: Type: text/plain, Size: 77 bytes --]


RMS: sorry for the duplicate; I had forgotten to send to the list as
well.


[-- Attachment #2.1: Type: text/plain, Size: 4605 bytes --]

Richard Stallman <rms@gnu.org> writes:

>     Sometime in the past year or so compilation-mode started acting very
>     strangely for me with a CVS-built emacs.  First, if I use next-error
>     while compilation is still going on, I will often, after a moment or
>     two, see the point jump unpredictably in the file it takes me to.
>
> Have you figured out any more about this?

Well, I haven't figured out much new, but it still happens.  I haven't
rebuilt Emacs since I first reported the problem, of course.

Here's some more detail.  I have a compilation that produces the
enclosed errors.  It's recompiling the same file and issuing errors
twice for it in this case because it's a bootstrap process, but I'm
positive that I've seen the behavior when errors only appear once.  If
I hit `f4' (bound to compilation-next-error for me) once, it goes to
the first error in the indicated source window.  Then when the 2nd
batch of errors show up in the compilation window, the point in the
open source window moves to the very last error in that window,
without any help from me.  The point in the *compilation* buffer
remains where it was.  

If I repeat the above process, but issue a build command that repeats
6 times, and then I hit `f4' *again* during compilation, the point
moves to the 2nd error in the source window and then after only a very
brief pause, jumps to the final error.

I haven't yet adjusted my compilation-error-regexp-alists to the new
scheme, so it's possible that has somethign to do with it.

>     I note that now there appears to be a two-layer scheme with
>     compilation-error-regexp-alist-alist that renders my old settings
>     invalid.  I can adjust what I'm doing, but discovering what to do was
>     a struggle.  Is there a way to provide a better transition experience
>     for people who upgrade?
>
> Can you tell us any more about where the difficulty was?

Well, my usual experience of being able to browse compilation errors
just stopped working, and I had to dig through the docs for
compilation-next-error to find out that the whole world was different.
It would have been nicer if the system had detected that I was using
the old variables and given me a warning about the change and pointed
me at the new docs, or better yet had done the translation
automatically for me.  It would also have been nicer if the new docs
included a transition guide.

Here's what my current (old) .emacs says about this stuff:

  ;;
  ;; Customize regular expressions which match compilation error messages
  ;;

  ;; Make sure we have a definition of compilation-error-regexp-alist before
  ;; modifying it
  (require 'compile)

  (setq compilation-error-regexp-alist
        (append
         '(
           ("\\s-*#\\s-*line\\s-+\\([0-9]+\\)\\s-+\"\\([a-zA-Z]?:?[^\":\n]+\\)\"" 2 1)

           ;; This expression can be used to match error messages from Jam
           ("\"?\\([a-zA-Z]?:?[^\":\n]+\\)\"?:[   ]+[Ll]ine[  ]+\\([0-9]+\\)" 1 2)

           ;; matches some gcc error output
           ("\\(\\(                 \\)\\|\\(In file included \\)\\)from \"?\\([a-zA-Z]?:?[^\":\n]+\\)\"?:\\([0-9]+\\)" 4 5)


           ;; matches some vc7 error output
           (" +\\(\\([a-zA-Z]:\\)?[^:(	\n]+\\)(\\([0-9]+\\)):.*'.*" 1 3)


           ;; This expression matches metrowerks command-line tool output. We just
           ;; pick up the line number, since the file name is handled below
           ("\\(### mw[^\n]+\n\\)?# +\\([0-9]+\\):" nil 2)

           ("        \\(\\([a-zA-Z]:\\)?[^:(\t\n]+\\)(\\([0-9]+\\)) \
  : see " 1 3)

           (".*at\\s-+[Ll]ine\\s-+\\([0-9]+\\)\\s-+of\\s-+\"\\([a-zA-Z]?:?[^\":\n]+\\)\"" 2 1)

           ; Python backtraces include these
           ("^[ \t]*File \"\\([a-zA-Z]?:?[^\":\n]+\\)\",[ \t]+[Ll]ine[ \t]+\\([0-9]+\\).*" 1 2)

           ;; These are "helpfully" supplied when Boost.Build testing prints a backtrace
           ("^[ \t]*\\(at\\|from\\)[ \t]+line[ \t]+\\([0-9]+\\)[ \t]+of[ \t]+\"?\\([a-zA-Z]?:?[^\":\n(]+\\).*" 3 2)
           )
         compilation-error-regexp-alist))

  (if (boundp 'compilation-file-regexp-alist)
      (setq compilation-file-regexp-alist
        (append
         '(
           ;; This expression matches metrowerks command-line tool output, which
           ;; only displays the filename once for multiple errors
           ("### mw[^\n]+\n# *In: \"?\\([a-zA-Z]?:?[^\":\n]+\\)\"?$" 1)
           ("### mw[^\n]+\n# *File: \"?\\([a-zA-Z]?:?[^\":\n]+\\)\"?$" 1)

           ("^(\\(.*\.tex\\)\\|\\(.*tex\.cfg\\)\nLaTeX2e" 1)
           )
             compilation-file-regexp-alist)))



[-- Attachment #2.2: error.log --]
[-- Type: text/plain, Size: 2984 bytes --]

-*- mode: compilation; default-directory: "c:/boost/tools/build/jam_src/" -*-
build
Setting environment for using Microsoft Visual Studio .NET 2003 tools.
(If you have another version of Visual Studio or Visual C++ installed and wish
to use its tools from the command line, run vcvars32.bat for that version.)
###
### Using 'vc7' toolset.
###

c:\boost\tools\build\jam_src>rd /S /Q bootstrap 

c:\boost\tools\build\jam_src>md bootstrap 

c:\boost\tools\build\jam_src>cl /nologo /GZ /Zi /MLd -DNT -DYYDEBUG kernel32.lib advapi32.lib /Febootstrap\jam0  command.c compile.c execnt.c execunix.c execvms.c expand.c filent.c fileos2.c fileunix.c filevms.c glob.c hash.c hdrmacro.c headers.c jam.c jambase.c jamgram.c lists.c make.c make1.c newstr.c option.c parse.c pathunix.c pathvms.c regexp.c rules.c scan.c search.c subst.c timestamp.c variable.c modules.c strings.c filesys.c builtins.c pwd.c class.c w32_getreg.c native.c modules/set.c modules/path.c modules/regex.c  modules/property-set.c modules/sequence.c modules/order.c 
command.c
compile.c
execnt.c
execnt.c(160) : warning C4090: 'function' : different 'const' qualifiers
execnt.c(171) : warning C4090: '=' : different 'const' qualifiers
execnt.c(181) : warning C4090: '=' : different 'const' qualifiers
execunix.c
execvms.c
expand.c
filent.c
fileos2.c
fileunix.c
filevms.c
glob.c
hash.c
hdrmacro.c
headers.c
jam.c
jambase.c
jamgram.c
lists.c
make.c
make1.c
Generating Code...
Compiling...
newstr.c
option.c
parse.c
pathunix.c
pathvms.c
regexp.c
rules.c
scan.c
search.c
subst.c
timestamp.c
variable.c
modules.c
strings.c
filesys.c
builtins.c
pwd.c
class.c
w32_getreg.c
native.c
Generating Code...
Compiling...
set.c
path.c
regex.c
property-set.c
sequence.c
order.c
Generating Code...

c:\boost\tools\build\jam_src>.\bootstrap\jam0 -f build.jam --toolset=vc7 "--toolset-root=C:\Program Files\Microsoft Visual Studio .NET 2003\Common7\Tools\\..\..\VC7\ " clean 
...found 1 target...
...updating 1 target...
[DELETE] clean
...updated 1 target...

c:\boost\tools\build\jam_src>.\bootstrap\jam0 -f build.jam --toolset=vc7 "--toolset-root=C:\Program Files\Microsoft Visual Studio .NET 2003\Common7\Tools\\..\..\VC7\ "          
...found 44 targets...
...updating 1 target...
[COMPILE] bin.ntx86\bjam.exe
command.c
compile.c
expand.c
glob.c
hash.c
hcache.c
headers.c
hdrmacro.c
jam.c
jambase.c
jamgram.c
lists.c
make.c
make1.c
newstr.c
option.c
parse.c
regexp.c
rules.c
scan.c
Generating Code...
Compiling...
search.c
subst.c
timestamp.c
variable.c
modules.c
strings.c
filesys.c
builtins.c
pwd.c
class.c
native.c
set.c
path.c
regex.c
property-set.c
sequence.c
order.c
execnt.c
execnt.c(160) : warning C4090: 'function' : different 'const' qualifiers
execnt.c(171) : warning C4090: '=' : different 'const' qualifiers
execnt.c(181) : warning C4090: '=' : different 'const' qualifiers
filent.c
pathunix.c
Generating Code...
Compiling...
w32_getreg.c
Generating Code...
...updated 1 target...

Compilation finished at Wed Jun 01 12:47:44

[-- Attachment #2.3: Type: text/plain, Size: 61 bytes --]


-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

[-- Attachment #3: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: "Recent" changes to compilation-mode?

[Richard Stallman]

    Well, my usual experience of being able to browse compilation errors
    just stopped working, and I had to dig through the docs for
    compilation-next-error to find out that the whole world was different.
    It would have been nicer if the system had detected that I was using
    the old variables and given me a warning about the change and pointed
    me at the new docs, or better yet had done the translation
    automatically for me.

I agree that something like this would be good.
I think something like this could be implemented as a superficial
change that would not alter the existing code, and thus would be safe.

Unfortunately I don't have time to do it myself.  Could someone please
study this and implement such a transition aid?

Re: "Recent" changes to compilation-mode?

[David Abrahams]

Richard Stallman <rms@gnu.org> writes:

>     If I repeat the above process, but issue a build command that repeats
>     6 times, and then I hit `f4' *again* during compilation, the point
>     moves to the 2nd error in the source window and then after only a very
>     brief pause, jumps to the final error.
>
> Is this result repeatable?  

Yes.

> If so, it should be possible to narrow down
> what code is doing this, just by inserting statements like
>
>   (push (list 'place-5 (window-start)) debug-list)
>
> in various places.  (Set debug-list to nil first.)

Sorry, you're well beyond the scope of my understanding of how to
debug emacs.  I'm happy to try whatever you like, but I think I need
some more guidance, e.g.

  a. Some examples of those various places

  b. How to see the results of my changes

  c. How to understand those results

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

Re: "Recent" changes to compilation-mode?

[Richard Stallman]

    >   (push (list 'place-5 (window-start)) debug-list)
    >
    > in various places.  (Set debug-list to nil first.)

    Sorry, you're well beyond the scope of my understanding of how to
    debug emacs.  I'm happy to try whatever you like, but I think I need
    some more guidance, e.g.

I am too overloaded to do this job just now.

Can someone else please give David this help?