error on start-up: font not defined

error on start-up: font not defined

[ken]

Just compiled emacs v.23.1.1 and, testing, ran src/emacs -q

Got the following error:

Font `---regular-R-Normal--15------ISO8859-1' is not defined

However, if I do

src/emacs -q -fn '-misc-fixed-medium-r-normal--12-*-*-*-c-90-iso8859-1'

and minor variants in the font specification, emacs loads okay.


What's the problem...?  Anybody know?

# uname -srvmpio
Linux 2.6.18-128.el5 #1 SMP Wed Jan 21 10:44:23 EST 2009 i686 i686 i386
GNU/Linux


Thanks.


-- 
War is a failure of the imagination.
        --William Blake

problem reading ~/.emacs.desktop [was: Re: error on start-up: font not defined]

[ken]

Okay, I found a work-around for the first problem, but ran into another:

In the process of a system upgrade, I pulled in last week's
~/.emacs.desktop.  And, yes, I upgraded emacs to v.23.1.1.  Now I get
this error when emacs loads:

Debugger entered--Lisp error: (void-variable Í)
  (desktop-create-buffer Í "/home/mine/file2open.html" "file2open.html"
(quote html-helper-mode) (quote (auto-fill-mode abbrev-mode)) 5535
(quote (nil nil)) nil nil (quote (...)))
  eval-buffer(#<buffer  *load*> nil "/home/zl/.emacs.desktop" nil t)  ;
Reading at buffer position 1139
  load-with-code-conversion("/home/zl/.emacs.desktop"
"/home/zl/.emacs.desktop" t t)
  load("/home/zl/.emacs.desktop" t t t)
  desktop-read()
  #[nil " \x18	ƒ\x0f


Anyone know how to fix this?  If so, much appreciated.



On 09/14/2009 06:57 PM ken wrote:
> Just compiled emacs v.23.1.1 and, testing, ran src/emacs -q
> 
> Got the following error:
> 
> Font `---regular-R-Normal--15------ISO8859-1' is not defined
> 
> However, if I do
> 
> src/emacs -q -fn '-misc-fixed-medium-r-normal--12-*-*-*-c-90-iso8859-1'
> 
> and minor variants in the font specification, emacs loads okay.
> 
> 
> What's the problem...?  Anybody know?
> 
> # uname -srvmpio
> Linux 2.6.18-128.el5 #1 SMP Wed Jan 21 10:44:23 EST 2009 i686 i686 i386
> GNU/Linux
> 
> 
> Thanks.
> 
> 

Re: problem reading ~/.emacs.desktop

[ken]

On 09/14/2009 11:09 PM ken wrote:
> Okay, I found a work-around for the first problem, but ran into another:
> 
> In the process of a system upgrade, I pulled in last week's
> ~/.emacs.desktop.  And, yes, I upgraded emacs to v.23.1.1.  Now I get
> this error when emacs loads:
> 
> Debugger entered--Lisp error: (void-variable Í)
>   (desktop-create-buffer Í "/home/mine/file2open.html" "file2open.html"
> (quote html-helper-mode) (quote (auto-fill-mode abbrev-mode)) 5535
> (quote (nil nil)) nil nil (quote (...)))
>   eval-buffer(#<buffer  *load*> nil "/home/zl/.emacs.desktop" nil t)  ;
> Reading at buffer position 1139
>   load-with-code-conversion("/home/zl/.emacs.desktop"
> "/home/zl/.emacs.desktop" t t)
>   load("/home/zl/.emacs.desktop" t t t)
>   desktop-read()
>   #[nil " \x18	ƒ\x0f
> 
> 
> Anyone know how to fix this?  If so, much appreciated.
> 

My (uneducated) hunch is that desktop-create-buffer doesn't like the
first argument.  This argument is a number (which in my email client is
displayed as an 'I' with a diacritical mark above it). When the above
Debugger output is displayed in emacs the number is '\315'.  If I look
at this ~/.emacs.desktop file with the "less" command, it is displayed
as "<81><CD>"... for every instance of a buffer.  (?)

According to the emacs doc on "desktop-create-buffer", this first arg is
"desktop-file-version"; I don't know what that is and have no idea why
the function "desktop-file-version" doesn't like it.

Anybody able to figure out what this problem is?

tia.


> 
> ....

Re: problem reading ~/.emacs.desktop

[Juanma Barranquero]

On Tue, Sep 15, 2009 at 11:24, ken <gebser@mousecar.com> wrote:

> My (uneducated) hunch is that desktop-create-buffer doesn't like the
> first argument.  This argument is a number (which in my email client is
> displayed as an 'I' with a diacritical mark above it). When the above
> Debugger output is displayed in emacs the number is '\315'.  If I look
> at this ~/.emacs.desktop file with the "less" command, it is displayed
> as "<81><CD>"... for every instance of a buffer.  (?)
>
> According to the emacs doc on "desktop-create-buffer", this first arg is
> "desktop-file-version"; I don't know what that is and have no idea why
> the function "desktop-file-version" doesn't like it.
>
> Anybody able to figure out what this problem is?

Can you edit the desktop file and change these Í to 205 (the literal
string "205", I mean)? The version number should be a literal, for
example:

(desktop-create-buffer 206
  "c:/my-file"
  ... etc)

Can you move aside your current desktop file and recreate one with
`desktop-save' (or `desktop-save-in-desktop-dir') to see whether it is
correctly created?

    Juanma

Re: problem reading ~/.emacs.desktop

[ken]

On 09/15/2009 05:37 AM Juanma Barranquero wrote:
> On Tue, Sep 15, 2009 at 11:24, ken <gebser@mousecar.com> wrote:
> 
>> My (uneducated) hunch is that desktop-create-buffer doesn't like the
>> first argument.  This argument is a number (which in my email client is
>> displayed as an 'I' with a diacritical mark above it). When the above
>> Debugger output is displayed in emacs the number is '\315'.  If I look
>> at this ~/.emacs.desktop file with the "less" command, it is displayed
>> as "<81><CD>"... for every instance of a buffer.  (?)
>>
>> According to the emacs doc on "desktop-create-buffer", this first arg is
>> "desktop-file-version"; I don't know what that is and have no idea why
>> the function "desktop-file-version" doesn't like it.
>>
>> Anybody able to figure out what this problem is?
> 
> Can you edit the desktop file and change these Í to 205 (the literal
> string "205", I mean)? The version number should be a literal, for
> example:
> 
> (desktop-create-buffer 206
>   "c:/my-file"
>   ... etc)
> 
> Can you move aside your current desktop file and recreate one with
> `desktop-save' (or `desktop-save-in-desktop-dir') to see whether it is
> correctly created?
> 
>     Juanma

Juanma,

Thanks much for replying.  You and I are thinking the same.  But emacs
isn't thinking like either of us.

The "desktop-file-version" argument *is* an integer... but in our email
clients it is being *displayed* as a character.  In emacs as I view the
old ".emacs.desktop" on my system (not in the Debugger, but as a normal
file/buffer), it displays as "205" (without the quotes).  The Linux
"less" command displays the same.

In the newly created ".emacs.desktop" file (created by the upgraded
version of emacs), this same argument is displayed as an 'I' with a
diacritical mark over it-- the same as the old one was in the Debugger.
  (?)  The value of that character (as determined by "C-x =") is 205.
This I don't understand: it's as if the Debugger incorrectly interpreted
the correct integer as a character and then complained about it.

Well, I used another workaround: I found that if I loaded the old
".emacs.desktop" into a buffer and then did "C-x e" at the end of each
"(desktop-create-buffer ...)", it would properly load each buffer....
Well, there were a couple files (out of over 100) which, for irrelevant
reasons, were absent from my system.  But none of the missing files were
what the Debugger complained about-- and the message which the Debugger
gave said nothing about a missing file, but rather indicated that the
first argument somehow was incorrect.  See my previous email.

Using the above workaround, I successfully load (nearly) all files
listed in ".emacs.desktop" into emacs buffers.  And I can see them all
using "C-x b" and I can visit them all.  But then I do a "desktop-save"
as you recommended, quit emacs, start emacs again, and get the same
Debugger error message again.

So then I edited ".emacs.desktop" to replace the 'I' with the
diacritical mark with "205" (without the quotes).  And I comment out the
code in ~/.emacs which invokes "desktop-read" and "desktop-save" and
reload emacs and run these by hand.  I find that emacs has changed "205"
to "206".  And though the files are read into emacs fine, I get this
different error from Debugger:

Debugger entered--Lisp error: (error "Local variables entry is missing
the suffix")
  signal(error ("Local variables entry is missing the suffix"))
  error("Local variables entry is missing the suffix")
  hack-local-variables()
  normal-mode(t)
  after-find-file(nil t)
  find-file-noselect-1(#<buffer g-mob> "~/pub/g-mob" nil nil
"~/pub/g-mob" (38245125 64771))
  find-file-noselect("/home/zl/pub/g-mob")
  desktop-restore-file-buffer("/home/zl/pub/g-mob" "g-mob" nil)
  #[nil "\b	A\b

This error isn't clear enough for me to figure out.  And it's probably
over the head of most, if not all, of the people on this list.  But if
anyone has a cogent guess as to how to fix this, I'd be willing to give
it a try.

Thanks again, Juanma, for your help.  It got me a little further toward
resolution.

Re: problem reading ~/.emacs.desktop

[Juanma Barranquero]

On Wed, Sep 16, 2009 at 10:14, ken <gebser@mousecar.com> wrote:

> In the newly created ".emacs.desktop" file (created by the upgraded
> version of emacs), this same argument is displayed as an 'I' with a
> diacritical mark over it-- the same as the old one was in the Debugger.
>  (?)  The value of that character (as determined by "C-x =") is 205.

?? With Emacs 23.1, the version should be 206.

I'd like to know what happens when you do the following

  - create an empty directory and cd to it
  - set HOME to that directory
  - create a test file (let's call it test.txt) with a couple lines
  - then
       emacs -Q test.txt
       M-x desktop-save <RET>

> So then I edited ".emacs.desktop" to replace the 'I' with the
> diacritical mark with "205" (without the quotes).  And I comment out the
> code in ~/.emacs which invokes "desktop-read" and "desktop-save" and
> reload emacs and run these by hand.  I find that emacs has changed "205"
> to "206".  And though the files are read into emacs fine, I get this
> different error from Debugger:
>
> Debugger entered--Lisp error: (error "Local variables entry is missing
> the suffix")
>  signal(error ("Local variables entry is missing the suffix"))
>  error("Local variables entry is missing the suffix")
>  hack-local-variables()
>  normal-mode(t)
>  after-find-file(nil t)
>  find-file-noselect-1(#<buffer g-mob> "~/pub/g-mob" nil nil
> "~/pub/g-mob" (38245125 64771))
>  find-file-noselect("/home/zl/pub/g-mob")
>  desktop-restore-file-buffer("/home/zl/pub/g-mob" "g-mob" nil)
>  #[nil "       A
>
> This error isn't clear enough for me to figure out.  And it's probably
> over the head of most, if not all, of the people on this list.  But if
> anyone has a cogent guess as to how to fix this, I'd be willing to give
> it a try.
>
> Thanks again, Juanma, for your help.  It got me a little further toward
> resolution.
>
>
>
>



-- 
    Juanma

Re: problem reading ~/.emacs.desktop

[Juanma Barranquero]

[sorry for the previous unfinished post]

On Wed, Sep 16, 2009 at 10:14, ken <gebser@mousecar.com> wrote:

> In the newly created ".emacs.desktop" file (created by the upgraded
> version of emacs), this same argument is displayed as an 'I' with a
> diacritical mark over it-- the same as the old one was in the Debugger.
>  (?)  The value of that character (as determined by "C-x =") is 205.

?? With Emacs 23.1, the version should be 206 in a newly created .desktop.

I'd like to know what happens when you do the following

 - create an empty directory and cd to it
 - set HOME to that directory
 - create a test file (let's call it test.txt) with a couple lines
 - then
      emacs -Q test.txt
      M-x desktop-save <RET> <RET>      ; to create a new desktop file
      C-x C-c
  followed by
      emacs -Q -f desktop-load

> So then I edited ".emacs.desktop" to replace the 'I' with the
> diacritical mark with "205" (without the quotes).  And I comment out the
> code in ~/.emacs which invokes "desktop-read" and "desktop-save" and
> reload emacs and run these by hand.

You shouldn't use `desktop-save' and `desktop-read' in you ~/.emacs, just

  (desktop-save-mode 1)

is enough.

> Debugger entered--Lisp error: (error "Local variables entry is missing
> the suffix")
>  signal(error ("Local variables entry is missing the suffix"))
>  error("Local variables entry is missing the suffix")

> This error isn't clear enough for me to figure out.

This is not related to desktop. Emacs is complaining about a problem
with the Local variables section in the file, i.e., the

  Local Variables:
  one-variable: whatever
  another-variable: blah blah
  End:

section near the end of the file.

Just out of curiosity: it is possible you're using CRLF files in a
context that does not expect them, or that some of the files have
mixed LF and CRLF line endings?

    Juanma

Re: problem reading ~/.emacs.desktop

[ken]

On 09/16/2009 04:39 AM Juanma Barranquero wrote:
> [sorry for the previous unfinished post]

That's quite alright.

> 
> On Wed, Sep 16, 2009 at 10:14, ken <gebser@mousecar.com> wrote:
> 
>> In the newly created ".emacs.desktop" file (created by the upgraded
>> version of emacs), this same argument is displayed as an 'I' with a
>> diacritical mark over it-- the same as the old one was in the Debugger.
>>  (?)  The value of that character (as determined by "C-x =") is 205.
> 
> ?? With Emacs 23.1, the version should be 206 in a newly created .desktop.
> 
> I'd like to know what happens when you do the following
> 
>  - create an empty directory and cd to it
>  - set HOME to that directory
>  - create a test file (let's call it test.txt) with a couple lines
>  - then
>       emacs -Q test.txt
>       M-x desktop-save <RET> <RET>      ; to create a new desktop file
>       C-x C-c
>   followed by
>       emacs -Q -f desktop-load

Very cool idea.  Here's what happened.

Created new directory, made it HOME, created a file there.

"emacs -Q test.txt" loaded fine.

"M-x desktop-save RET RET" did create a new "~/.emacs.desktop".  Here it is:

============================ start ================================
;; -*- mode: emacs-lisp; coding: emacs-mule; -*-
;;
--------------------------------------------------------------------------
;; Desktop File for Emacs
;;
--------------------------------------------------------------------------
;; Created Sat Sep 19 20:02:35 2009
;; Desktop file format version 206
;; Emacs version 23.1.1

;; Global section:
(setq desktop-missing-file-warning nil)
(setq tags-file-name nil)
(setq tags-table-list nil)
(setq search-ring nil)
(setq regexp-search-ring nil)
(setq register-alist nil)

;; Buffer section -- buffers listed in same order as in buffer list:
(desktop-create-buffer 206
  "/home/zl/emtmp/test.txt"
  "test.txt"
  'text-mode
  nil
  1
  '(nil nil)
  nil
  nil
  '((buffer-file-coding-system . undecided-unix)))

=========================== end ========================

[Note: I displayed the above from the shell, not in emacs (just to keep
with the test.  Also worth noting: the version is *not* 205, but 206.]

After exiting emacs and going back in with "emacs -Q -f desktop-load", I
got the error message in the minibuffer

Symbol's function definition is void: desktop-load

and so test.txt wasn't loaded, "C-x C-b" showed just

------------------------ start -----------------------
.   *scratch*                0  Lisp Interaction
  * *Messages*             132  Fundamental
------------------------- end ------------------------

If, however, I do "M-x desktop-read" test.txt is loaded as expected.




> 
>> So then I edited ".emacs.desktop" to replace the 'I' with the
>> diacritical mark with "205" (without the quotes).  And I comment out the
>> code in ~/.emacs which invokes "desktop-read" and "desktop-save" and
>> reload emacs and run these by hand.
> 
> You shouldn't use `desktop-save' and `desktop-read' in you ~/.emacs, just
> 
>   (desktop-save-mode 1)
> 
> is enough.

Great!  That's a pile of code I can pull out of my .emacs from years back.


> 
>> Debugger entered--Lisp error: (error "Local variables entry is missing
>> the suffix")
>>  signal(error ("Local variables entry is missing the suffix"))
>>  error("Local variables entry is missing the suffix")
> 
>> This error isn't clear enough for me to figure out.
> 
> This is not related to desktop. Emacs is complaining about a problem
> with the Local variables section in the file, i.e., the
> 
>   Local Variables:
>   one-variable: whatever
>   another-variable: blah blah
>   End:
> 
> section near the end of the file.

Hey, you're good.  Yeah, I've got some of that in one of my files.  But
what's the "suffix" that's missing?


> 
> Just out of curiosity: it is possible you're using CRLF files in a
> context that does not expect them, or that some of the files have
> mixed LF and CRLF line endings?

Not likely, but, yeah, possible.  Occasionally I'll work on a file that
came from Windows.  Do you mean an emacs configuration file?  Given the
results above, is it relevant?


Juanma, thanks very much for your help.
ken

> 
>     Juanma

Re: problem reading ~/.emacs.desktop

[Juanma Barranquero]

On Sun, Sep 20, 2009 at 05:04, ken <gebser@mousecar.com> wrote:

> Very cool idea.  Here's what happened.

> After exiting emacs and going back in with "emacs -Q -f desktop-load", I
> got the error message in the minibuffer
>
> Symbol's function definition is void: desktop-load

> If, however, I do "M-x desktop-read" test.txt is loaded as expected.

Yeah, my mistake, I meant desktop-read.

This test shows that your Emacs setup does not have any trouble
reading or writing desktop files. So the one that gives you trouble
must have some formatting problem or somesuch. I'd suggest just
deleting it and recreating it (by loading the files you want to have
and then M-x desktop-save'ing).

> Hey, you're good.  Yeah, I've got some of that in one of my files.  But
> what's the "suffix" that's missing?

It means that Emacs thinks that the Local Variables section has a
format problem; that's why I ask about CRLF vs LF.

The Local Variables section is a bit strict about formatting. Entries
can have prefixes, for example:

-- Local Variables:
-- mode: text
-- tab-width: 4
-- End:

If the Local Variables line has a prefix, all of them must too.
Similarly, if there's a suffix:

-- Local Variables: ==
-- mode: text       ==
-- tab-width: 5     ==
-- End:             ==

will work, but

-- Local Variables: ==
-- mode: text       ==
-- tab-width: 5     =
-- End:             ==

will produce the "missing suffix" error.

> Not likely, but, yeah, possible.  Occasionally I'll work on a file that
> came from Windows.  Do you mean an emacs configuration file?  Given the
> results above, is it relevant?

It could be relevant in the Local Variables problem.

    Juanma

Re: problem reading ~/.emacs.desktop

[ken]

On 09/19/2009 11:29 PM Juanma Barranquero wrote:
> On Sun, Sep 20, 2009 at 05:04, ken <gebser@mousecar.com> wrote:
> 
>> Very cool idea.  ....

I like the idea of creating, in effect, a new and totally clean
environment in which to test.  Another step in the process to implement:
before setting the new $HOME variable, run "bash -" (or whatever shell
used).  This creates a new shell.  The new shell's environment, though a
copy of the previous, can be manipulated and changed and after testing
easily wiped out simply by exiting that new shell.


> 
>> ....
>
> Yeah, my mistake, I meant desktop-read.

Small thing.  Funny how the mind can lose one little piece in a long
series of steps.  We were already by then on the same page, so no big deal.


> 
> This test shows that your Emacs setup does not have any trouble
> reading or writing desktop files. So the one that gives you trouble
> must have some formatting problem or somesuch. I'd suggest just
> deleting it and recreating it (by loading the files you want to have
> and then M-x desktop-save'ing).

In fact that's pretty much what I did.  I grepped the old
".emacs.desktop" to output just the filenames and fed this into emacs at
load time.  I knew emacs would load all the files, but I didn't think it
would remember the point (text cursor) position was in my previous
session.  Somehow it did though... that's a puzzle to me.

That was my plan of last resort-- at least I would have the files loaded
that I wanted.  My other concern though was for other folks who upgrade
emacs and would expect it to load up all their files from their last
pre-upgrade session; this wouldn't happen for them.  That is,
.emacs.desktop has (I think) a version number; it should recognize it
and take appropriate action and provide a smoother upgrade path.  This
would include an accounting for a simultaneous system upgrade from some
other locale definitions to UTF-8; this might have been (part of?) the
problem with "205" displaying as an 'I' with a diacritical mark.  I'm
not sure... just a hunch.



> 
>> Hey, you're good.  Yeah, I've got some of that in one of my files.  But
>> what's the "suffix" that's missing?
> 
> It means that Emacs thinks that the Local Variables section has a
> format problem; that's why I ask about CRLF vs LF.
> 
> ....

Thanks for the explanation... very good and very clear.  With that I'll
search through my files for Local Variables and should be able to locate
and fix what needs it.

Juanma, thanks for all the tips, help, and clarifications.  You're on my
A-Team list.


Best,
ken