Re: [coreutils] RFC for adding file creation mode feature into touch utility.

[Jim Meyering]

Rakib Mullick wrote:
>> I confess that my first reaction was "why bother?"
>> Touch is nominally a very simple tool, and adding --mode support
>> feels like adding unwarranted bloat -- and not in line with
>> the one-job-one-tool philosophy.
>
> Yes, right. Then, touch only supposed to change file's timestamp. No?
> But, touch also creates a file for us. Not only it creates a file for
> us, when it creates a file it takes reference from other files. So,
> althrough we're in philosophy of one-tool-do-one-job-well, there are
> something else that we can get.
>
> And if adding --mode option to touch gets really out of philosophy,
> then I think we need a new tool to for creating file with all the
> options that we might have want. That will truly help us to keep the
> philosophy intact.

?

When the utility of a new feature is not obvious, the proposal for
that new feature must be accompanied by significant justification,
and so far, the argument that "adding this option to touch saves us the
trouble of invoking chmod" is insufficient.

Typical "good" justification usually goes like this:

    I want to do X using POSIX tools, and the portable way to do it is
    too cumbersome, but if I add this option to one of the coreutils,
    it becomes much easier.  Here is how to do X using current tools:

        ...list of actual commands (inefficient, cumbersome, etc)...

    and now, do X using the program with the proposed change:

        ...list of commands demonstrating benefits of proposed change...

Then, if there is no reasonably simple/concise way to solve the problem
with some other tool, we evaluate whether some coreutils program
should be extended to do the job.

>> Why would one prefer the GNU-specific touch --mode ...
>> over the portable combination of touch and chmod?
>>
> Cause its easier to do 'touch --mode xxx filename' than 'touch
> filename' and 'chmod xxx'.

As mentioned above, that is not enough justification.

> And Eric also pointed out that we will be
> able to do 'touch -r a -M b' to create a referenced file with
> different mode.

Let's back up a step...
Why do you want an option for this when you can already use touch
to create a file with selected permissions via your umask?
If you say that you require an execute bit or some special bit to
be set, please explain why.  I do not see how having such bits set
on an empty file would be useful.  If you're about to add content,
then the time stamps set by touch will be overwritten, and you might as
well have created the file via the shell, (if there was only one) say
with ": > FILE".