VC shows stderr in target buffer

VC shows stderr in target buffer

[David Reitter]

VC shows warnings from CVS in the target buffer, e.g. when invoking C- 
x C-v =, I would get things like

cvs diff: warning: failed to open /Users/dr/.cvspass for reading: No  
such file or directory

in the *vc-diff* buffer.  Wouldn't this be better shown in the echo  
area, if at all?
The warnings are quite a pain in the buffer, given that I would  
normally copy&paste the whole contents from there.

A quick look at vc.el suggests that `vc-do-command' calls `start- 
process', directing stdout and stderr into the current buffer. 
  

Re: VC shows stderr in target buffer

[Stefan Monnier]

> VC shows warnings from CVS in the target buffer, e.g. when invoking C-
> x C-v =, I would get things like

> cvs diff: warning: failed to open /Users/dr/.cvspass for reading: No such
> file or directory

> in the *vc-diff* buffer.  Wouldn't this be better shown in the echo area, if
> at all?

It seems better in the *vc-diff* buffer where the user will be sure to
see it than in the echo area where it'll be overwritten soon after
(especially for multi-line messages).

> The warnings are quite a pain in the buffer, given that I would normally
> copy&paste the whole contents from there.

Any reason why you can't fix the problem so that CVS doesn't complain?

> A quick look at vc.el suggests that `vc-do-command' calls `start-
> process', directing stdout and stderr into the current buffer.

Yes, it's done on purpose because it seems like the best place to put
this info.


        Stefan

Re: VC shows stderr in target buffer

[David Reitter]

On 4 Apr 2008, at 15:09, Stefan Monnier wrote:
>
> It seems better in the *vc-diff* buffer where the user will be sure to
> see it than in the echo area where it'll be overwritten soon after
> (especially for multi-line messages).

When you do C-x C-s and the file can't be written, do you insert an  
error message into the buffer?
When you handle an exception signalled anywhere, do you insert an  
error message into the buffer or do you show it in the echo area?

My request was about consistency.  It's probably important to know I  
would do things like C-x C-v = in order to do something else with the  
resulting diff than to just read the output.  Thus, warnings (and they  
are just warnings!) are annoying when they're shown in the buffer.


> Any reason why you can't fix the problem so that CVS doesn't complain?

Sure, I can fix my problem, but that doesn't mean it's right to mix  
error messages and actual output...

Re: VC shows stderr in target buffer

[Stefan Monnier]

>> It seems better in the *vc-diff* buffer where the user will be sure to
>> see it than in the echo area where it'll be overwritten soon after
>> (especially for multi-line messages).

> When you do C-x C-s and the file can't be written, do you insert an error
> message into the buffer?
> When you handle an exception signalled anywhere, do you insert an error
> message into the buffer or do you show it in the echo area?

> My request was about consistency.

I do not think consistency in this regard is important.

> It's probably important to know I would
> do things like C-x C-v = in order to do something else with the  resulting
> diff than to just read the output.  Thus, warnings (and they  are just
> warnings!) are annoying when they're shown in the buffer.

The content of the diff output already often includes various extra
lines that can be more or less relevant depending on the tool used.
Some of those lines will be on stderr, others on stdout.  Tools that use
patches know to ignore leading and trailing garbage, so I really don't
think it's that important.


        Stefan