On Fri, May 7, 2010 at 9:05 AM, James Youngman <[email protected]> wrote:
> On Thu, May 6, 2010 at 6:32 PM, Eric Blake <[email protected]> wrote:
>> On 05/06/2010 09:13 AM, Ludovic Courtès wrote:
>>>> Even if it is also available as a shell builtin, that does not
>>>> alleviate the requirement to also have it available on $PATH.
>>>
>>> echo(1) *is* available in $PATH in the build environment, just like all
>>> of Coreutils.
>>>
>>> What would you think of ‘--with-echo’?
>>
>> Why is xargs blindly calling /bin/echo?
>
> I don't know, it's been that way since the first checkin in the
> (current) repository, dating from 1996.
>
>
>> Wouldn't it just be simpler to
>> change xargs.c' definition of default_cmd to be "echo" instead of
>> "/bin/echo" and using execvp instead of execve?
>
> I guess there will be a slight performance penalty, but if we want to
> prevent that we have the option of searching the path for the binary
> and using the absolute path.
>
>
>> In a compliant
>> environment, you are guaranteed that echo(1) is on PATH, and calling the
>> wrong echo because of a non-compliant environment is not really xargs'
>> concern.  In other words, instead of making it more complex by adding
>> configure --with-echo, I propose making it simpler by dropping the
>> hard-coding in the first place.
>
> I guess it makes sense.   Would you like to go ahead and make the change?

I just got another Hydra fail notification because of this issue, so
I'll go ahead and fix it.

James.

Reply via email to