bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

Please write in English if possible, because the Emacs maintainers
usually do not have translators to read other languages for them.

Your bug report will be posted to the emacs-pretest-bug@gnu.org mailing list.

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:

In an Rmail buffer, type "C-x C-s" to save it: you will almost never
see the usual "Saving file FOO..." message, because buffer-soze,
called by save-buffer, returns the size of the current message, not of
the entire message collection.

If Emacs crashed, and you have the Emacs process in the gdb debugger,
please include the output from the following gdb commands:
    `bt full' and `xbacktrace'.
If you would like to further debug the crash, please read the file
d:/gnu/emacs/etc/DEBUG for instructions.


In GNU Emacs 23.0.60.1 (i386-mingw-nt5.1.2600)
 of 2009-01-31 on HOME-C4E4A596F7
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4)'

Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: ENU
  value of $XMODIFIERS: nil
  locale-coding-system: cp1255
  default-enable-multibyte-characters: t

Major mode: RMAIL

Minor modes in effect:
  tooltip-mode: t
  tool-bar-mode: t
  mouse-wheel-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  global-auto-composition-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t

Recent input:
M-x l o a f d <backspace> <backspace> d - f i <tab> 
<return> l i s <tab> / f i l e s . e l <return> C-x 
C-f <up> <return> C-s s a v i n g SPC <help-echo> <help-echo> 
<help-echo> f i l e SPC <up> <up> <down> <down> C-s 
C-s C-r C-r C-r <up> M-x e d e b u <tab> d e <tab> 
<return> C-u M-x r m a i l <return> ~ / d a <tab> I 
N B O X . n e w <tab> <return> n p o d a <tab> f o 
o 1 2 3 4 5 . m b b <return> y e s <return> C-x C-s 
SPC SPC SPC SPC SPC SPC SPC G M-x r e p o r t <tab> 
<return>

Recent messages:
Added to d:/usr/eli/data/foo12345.mbb
 [2 times]
Result: t
 [3 times]
Result: 5470 (#o12536, #x155e, ?ᕞ)

Result: nil

Go-Nonstop...
Wrote d:/usr/eli/data/INBOX.new

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Stefan Monnier]

>     In an Rmail buffer, type "C-x C-s" to save it: you will almost never
>     see the usual "Saving file FOO..." message, because buffer-soze,
>     called by save-buffer, returns the size of the current message, not of
>     the entire message collection.

> The clean way to fix this is with something like `buffer-swapped-with'.

> I did not remember the existence of this feature when I recommended
> implementing `buffer-swapped-with', so I can't really say "I told you
> so."  But there is a general tendency that these sorts of features
> need to be checked and acted on in various places in the code, and
> most of them don't have any hooks.

This problem at least simply points to a current bug in save-buffer.
Maybe with buffer-swapped-with you could work around the bug rather than
fixing it.  Or even fail to see that it's a general bug rather than
a problem with buffer-swapped-with.  I'm not sure it's a benefit.


        Stefan

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: Eli Zaretskii <eliz@gnu.org>,  2137@emacsbugs.donarmstrong.com,  emacs-devel@gnu.org
> Date: Sun, 01 Feb 2009 20:35:12 -0500
> 
> This problem at least simply points to a current bug in save-buffer.

Which bug is that?

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Glenn Morris]

Eli Zaretskii wrote:

>> This problem at least simply points to a current bug in save-buffer.
>
> Which bug is that?

Well, one thing I wonder about is: how does it know the correct buffer
_contents_, but not the correct buffer _size_...

(Ie, why isn't it overwriting your mailbox file with just the current
message you happen to be viewing at the time?)

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Richard M Stallman]

    This problem at least simply points to a current bug in save-buffer.

It is a bug, for certain.  What I'm saying is that it is easy to fix
with something like `buffer-swapped-with', and hard to fix otherwise.

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Cc: Stefan Monnier <monnier@iro.umontreal.ca>,  rms@gnu.org,  2137@emacsbugs.donarmstrong.com
> Date: Mon, 02 Feb 2009 23:12:23 -0500
> 
> Eli Zaretskii wrote:
> 
> >> This problem at least simply points to a current bug in save-buffer.
> >
> > Which bug is that?
> 
> Well, one thing I wonder about is: how does it know the correct buffer
> _contents_, but not the correct buffer _size_...
> 
> (Ie, why isn't it overwriting your mailbox file with just the current
> message you happen to be viewing at the time?)

See rmail-write-region-annotate: it switches to the right buffer just
in time for write-region to save the correct text, but buffer-size is
called before that.

IMO, using annotations for the job of saving the mailbox is also a
horrible kludge.  All that just to avoid defining an Rmail-specific
binding for C-x C-s ...

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Richard M Stallman <rms@gnu.org>
> CC: eliz@gnu.org, 2137@emacsbugs.donarmstrong.com, emacs-devel@gnu.org
> Date: Tue, 03 Feb 2009 04:59:36 -0500
> 
>     This problem at least simply points to a current bug in save-buffer.
> 
> It is a bug, for certain.  What I'm saying is that it is easy to fix
> with something like `buffer-swapped-with', and hard to fix otherwise.

Okay, I give up: what (or who) is `buffer-swapped-with', and how is it
related to this issue?

As for an alternative for fixing this, if the current kludgey way of
using annotations for switching to the right buffer behind
save-buffer's back is acceptable, we could add to
rmail-write-region-annotate another hack: a call to buffer-size
followed by the same message "Saving file ..." that should have been
displayed by save-buffer.

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Stefan Monnier]

> IMO, using annotations for the job of saving the mailbox is also a
> horrible kludge.  All that just to avoid defining an Rmail-specific
> binding for C-x C-s ...

Thanks for your support,


        Stefan

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Stefan Monnier]

>> This problem at least simply points to a current bug in save-buffer.
> Which bug is that?

The bug is that is uses buffer-size to determine the size of the file,
without taking into account the fact that
write-region-annotate-functions can add lot of stuff to the file that's
not in the buffer's text.

The simplest fix is to simply always output the message rather than only
for files larger than 50KB.
The more difficult one is to move the message from save-buffer to
write-region.


        Stefan

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Stefan Monnier <monnier@IRO.UMontreal.CA>
> Cc: rms@gnu.org, 2137@emacsbugs.donarmstrong.com, emacs-devel@gnu.org
> Date: Tue, 03 Feb 2009 16:29:08 -0500
> 
> >> This problem at least simply points to a current bug in save-buffer.
> > Which bug is that?
> 
> The bug is that is uses buffer-size to determine the size of the file,
> without taking into account the fact that
> write-region-annotate-functions can add lot of stuff to the file that's
> not in the buffer's text.

How can buffer-size take that into account?  The correct full size of
what's written to the disk file is never seen, since write-region does
its job piecemeal, and the annotations are applied separately to each
piece by a_write.

Am I missing something?

> The simplest fix is to simply always output the message rather than only
> for files larger than 50KB.

That may be the only (non-kludgey) solution.

> The more difficult one is to move the message from save-buffer to
> write-region.

If I'm right above, this won't solve the case where annotations add a
lot.

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Stefan Monnier <monnier@IRO.UMontreal.CA>
> Cc: Glenn Morris <rgm@gnu.org>, rms@gnu.org, 2137@emacsbugs.donarmstrong.com
> Date: Tue, 03 Feb 2009 16:26:21 -0500
> 
> > IMO, using annotations for the job of saving the mailbox is also a
> > horrible kludge.  All that just to avoid defining an Rmail-specific
> > binding for C-x C-s ...
> 
> Thanks for your support,

Btw, isn't there's one more misfeature with write-region due to
rmail-write-region-annotate: if you use write-region to save a portion
of the current message to a file, would Rmail do that from the message
collection buffer instead? if so, what would be the encoding of
non-ASCII characters in the region as saved to the file?

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Richard M Stallman]

    > IMO, using annotations for the job of saving the mailbox is also a
    > horrible kludge.  All that just to avoid defining an Rmail-specific
    > binding for C-x C-s ...

    Thanks for your support,

An Rmail-specific binding for C-x C-s is not enough to make saving
work correctly.  Users can save the file through C-x s also, and other
commands too.  The common mechanism for these commands is
`save-buffer'; whatever makes them save the right text has to work
inside there.

Eli wrote:

     we could add to
    rmail-write-region-annotate another hack: a call to buffer-size
    followed by the same message "Saving file ..." that should have been
    displayed by save-buffer.

The difficulty here is to display that message only in the case
where the hook was ultimately called from `save-buffer'.  This could
be done by making `save-buffer' bind some variable for such hooks
to check.

However, such fixes are less clean that implementing `buffer-swapped-with'
and checking it in the necessary places.  I previously implemented it in
`basic-save-buffer', but just moving the implementation to `save-buffer'
would make it DTRT for this.  The fix would add no complexity.

    The simplest fix is to simply always output the message rather than only
    for files larger than 50KB.

That would be easy, but not quite the most convenient behavior.

    The more difficult one is to move the message from save-buffer to
    write-region.

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Eli Zaretskii <eliz@gnu.org>
> Cc: rms@gnu.org, 2137@emacsbugs.donarmstrong.com
> 
> Btw, isn't there's one more misfeature with write-region due to
> rmail-write-region-annotate: if you use write-region to save a portion
> of the current message to a file, would Rmail do that from the message
> collection buffer instead?

To answer myself: no, rmail-write-region-annotate does nothing if
write-region is invoked with its START argument non-nil (which means
to write only a portion of the current buffer).  So in this case, the
current message will be written to the file using
buffer-file-coding-system, as the user should expect.

Sorry for the noise.

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Stefan Monnier]

>> The bug is that is uses buffer-size to determine the size of the file,
>> without taking into account the fact that
>> write-region-annotate-functions can add lot of stuff to the file that's
>> not in the buffer's text.
> How can buffer-size take that into account?

It can't.  Luckily we don't have to use buffer-size for that.

> The correct full size of what's written to the disk file is never
> seen, since write-region does its job piecemeal, and the annotations
> are applied separately to each piece by a_write.

Indeed.

> Am I missing something?

We do have all the annotations in a single place, so we can compute the
final total size before we start writing.

>> The simplest fix is to simply always output the message rather than only
>> for files larger than 50KB.
> That may be the only (non-kludgey) solution.

Yes, I find it to be the best solution.  I see no benefit in avoiding
the message for smaller files.


        Stefan

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Stefan Monnier]

> An Rmail-specific binding for C-x C-s is not enough to make saving
> work correctly.  Users can save the file through C-x s also, and other
> commands too.  The common mechanism for these commands is
> `save-buffer'; whatever makes them save the right text has to work
> inside there.

save-buffer is not enough either in order to handle `autosave' (which
was the original reason for my using write-region-annotate-functions
rather than write-contents-functions).

>     The simplest fix is to simply always output the message rather than only
>     for files larger than 50KB.
> That would be easy, but not quite the most convenient behavior.

What would be the harm?


        Stefan

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Richard M Stallman <rms@gnu.org>
> CC: eliz@gnu.org, rgm@gnu.org, 2137@emacsbugs.donarmstrong.com
> Date: Wed, 04 Feb 2009 13:27:49 -0500
> 
>     > IMO, using annotations for the job of saving the mailbox is also a
>     > horrible kludge.  All that just to avoid defining an Rmail-specific
>     > binding for C-x C-s ...
> 
>     Thanks for your support,
> 
> An Rmail-specific binding for C-x C-s is not enough to make saving
> work correctly.  Users can save the file through C-x s also, and other
> commands too.  The common mechanism for these commands is
> `save-buffer'; whatever makes them save the right text has to work
> inside there.

So we could introduce some mechanism specific for such situation, like
some hook that save-buffer would call or something.  Annotations just
aren't that mechanism, they exist to add text to the buffer before it
is written.

>      we could add to
>     rmail-write-region-annotate another hack: a call to buffer-size
>     followed by the same message "Saving file ..." that should have been
>     displayed by save-buffer.
> 
> The difficulty here is to display that message only in the case
> where the hook was ultimately called from `save-buffer'.

rmail-write-region-annotate uses its first argument to deduce that it
is being called to write the entire buffer.  Why couldn't the same
logic be used for the message?

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Eli Zaretskii]

> From: Stefan Monnier <monnier@IRO.UMontreal.CA>
> Cc: eliz@gnu.org, rgm@gnu.org, 2137@emacsbugs.donarmstrong.com
> Date: Wed, 04 Feb 2009 14:49:40 -0500
> 
> > An Rmail-specific binding for C-x C-s is not enough to make saving
> > work correctly.  Users can save the file through C-x s also, and other
> > commands too.  The common mechanism for these commands is
> > `save-buffer'; whatever makes them save the right text has to work
> > inside there.
> 
> save-buffer is not enough either in order to handle `autosave'

Why?  There's no need to autosave the buffer in which we show
messages, and the buffer where the whole collection is held, if it's
modified, will be autosaved by the usual machinery.  What am I
missing?

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Stefan Monnier]

> Why?  There's no need to autosave the buffer in which we show
> messages, and the buffer where the whole collection is held, if it's
> modified, will be autosaved by the usual machinery.  What am I
> missing?

The buffer where the `mbox' is held doesn't have a buffer-file-name
(the view buffer has buffer-file-name set instead, so without
write-region-annotate-functions, the autosave file contains the
currently shown message rather than a copy of the mbox).


        Stefan

bug#2137: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message

[Richard M Stallman]

    save-buffer is not enough either in order to handle `autosave' (which
    was the original reason for my using write-region-annotate-functions
    rather than write-contents-functions).

That is true -- the implementation of saving for Rmail buffers needs
to handle auto-save one way or another.  That doesn't necessarily mean
it has to work through this hook.

    >     The simplest fix is to simply always output the message rather than only
    >     for files larger than 50KB.
    > That would be easy, but not quite the most convenient behavior.

    What would be the harm?

Nowadays, I guess it is ok to display the message for all files.

bug#2137: marked as done (23.0.60; Saving Rmail buffer does not show the "Saving file ..." message)

[Emacs bug Tracking System]

[-- Attachment #1: Type: text/plain, Size: 935 bytes --]


Your message dated Sat, 07 Feb 2009 12:24:43 +0200
with message-id <ud4dutuhg.fsf@gnu.org>
and subject line Re: bug#2137: 23.0.60;	Saving Rmail buffer does not show the "Saving file ..." message
has caused the Emacs bug report #2137,
regarding 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@emacsbugs.donarmstrong.com
immediately.)


-- 
2137: http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=2137
Emacs Bug Tracking System
Contact owner@emacsbugs.donarmstrong.com with problems

[-- Attachment #2: Type: message/rfc822, Size: 4583 bytes --]

From: Eli Zaretskii <eliz@gnu.org>
To: emacs-pretest-bug@gnu.org
Subject: 23.0.60; Saving Rmail buffer does not show the "Saving file ..." message
Date: Sat, 31 Jan 2009 21:13:03 +0200
Message-ID: <uab97e1c0.fsf@gnu.org>


Please write in English if possible, because the Emacs maintainers
usually do not have translators to read other languages for them.

Your bug report will be posted to the emacs-pretest-bug@gnu.org mailing list.

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:

In an Rmail buffer, type "C-x C-s" to save it: you will almost never
see the usual "Saving file FOO..." message, because buffer-soze,
called by save-buffer, returns the size of the current message, not of
the entire message collection.

If Emacs crashed, and you have the Emacs process in the gdb debugger,
please include the output from the following gdb commands:
    `bt full' and `xbacktrace'.
If you would like to further debug the crash, please read the file
d:/gnu/emacs/etc/DEBUG for instructions.


In GNU Emacs 23.0.60.1 (i386-mingw-nt5.1.2600)
 of 2009-01-31 on HOME-C4E4A596F7
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4)'

Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: ENU
  value of $XMODIFIERS: nil
  locale-coding-system: cp1255
  default-enable-multibyte-characters: t

Major mode: RMAIL

Minor modes in effect:
  tooltip-mode: t
  tool-bar-mode: t
  mouse-wheel-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  global-auto-composition-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t

Recent input:
M-x l o a f d <backspace> <backspace> d - f i <tab> 
<return> l i s <tab> / f i l e s . e l <return> C-x 
C-f <up> <return> C-s s a v i n g SPC <help-echo> <help-echo> 
<help-echo> f i l e SPC <up> <up> <down> <down> C-s 
C-s C-r C-r C-r <up> M-x e d e b u <tab> d e <tab> 
<return> C-u M-x r m a i l <return> ~ / d a <tab> I 
N B O X . n e w <tab> <return> n p o d a <tab> f o 
o 1 2 3 4 5 . m b b <return> y e s <return> C-x C-s 
SPC SPC SPC SPC SPC SPC SPC G M-x r e p o r t <tab> 
<return>

Recent messages:
Added to d:/usr/eli/data/foo12345.mbb
 [2 times]
Result: t
 [3 times]
Result: 5470 (#o12536, #x155e, ?ᕞ)

Result: nil

Go-Nonstop...
Wrote d:/usr/eli/data/INBOX.new




[-- Attachment #3: Type: message/rfc822, Size: 1976 bytes --]

From: Eli Zaretskii <eliz@gnu.org>
To: rms@gnu.org, 2137-done@emacsbugs.donarmstrong.com
Cc: monnier@IRO.UMontreal.CA
Subject: Re: bug#2137: 23.0.60;	Saving Rmail buffer does not show the "Saving file ..." message
Date: Sat, 07 Feb 2009 12:24:43 +0200
Message-ID: <ud4dutuhg.fsf@gnu.org>

> Date: Thu, 05 Feb 2009 11:39:39 -0500
> Cc: 2137@emacsbugs.donarmstrong.com
> 
>     >     The simplest fix is to simply always output the message rather than only
>     >     for files larger than 50KB.
>     > That would be easy, but not quite the most convenient behavior.
> 
>     What would be the harm?
> 
> Nowadays, I guess it is ok to display the message for all files.

That's what I did.