walt <w41...@gmail.com> writes:

> On 04/09/2010 08:19 AM, Harry Putnam wrote:
>> This is not a question about cvs... its only used for example.
>>
>> I'm puzzled about a change in what I see when I run
>>     cvs -n update 2>  /dev/null
>>
>> I've apparently lost the ability to remove stder from output.
>>
>> I used that command to trim out file descriptor 2 which used to leave
>> a list of any changed files in the repo on the console, for a very
>> long time.
>>
>>
>> Suddenly there is no difference with:
>>
>>        cvs -n update 2>  /dev/null
>>           cvs -n update
>>
>> The stuff on stderr still shows in the ouput either way.
>>
>> Further;
>>    cvs -n update 2>er  (redirect stder to ./er)
>>
>> Doesn't put anything in ./er
>>
>> However cvs -n update 1>out   (redirect stdout to ./out)
>>
>> Does catch the output I'm after and leave out stderr. (as one would
>> expect)
>>
>> So, again, apparently I've lost the ability to trim out stderr with a
>> redirect to /dev/null (cvs -n update 2>  /dev/null)
>>
>> -------        ---------       ---=---       ---------      --------
>>
>> The only thing I've been tinkering with is evaluating the
>> /etc/DIR_COLORS file.  I switched from evaluating a custom version to
>> evaluating the default version.
>
> I have no helpful advice, but I would try a couple of simple experiments:
>
> I have this in my home directory because I'm color blind:
> -rw-r--r--  1 wa1ter users       0 2007-08-27 18:29 .dir_colors
>
> $cat nonexistantfile
> cat: nonexistantfile: No such file or directory
>
> $cat nonexistantfile 2> /tmp/testfile
> $
>
> $cat /tmp/testfile
> cat: nonexistantfile: No such file or directory

Thanks... 

> Are you running cvs as root, or user, or ...?

I was running cvs as user, and now trying your tests... it appears the
trouble has stopped... doesn't occur now in cvs cmds either.

There was a reboot in between, so may never now what was going on.

Prior to rebooting I had tried to get a fresh env by ssh
u...@localhost from an xterm.  Hoping to rule out some oddball env
problem, but the file descriptor problem persisted.  However it has
apparently not survived a reboot.


Reply via email to