guix-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 02/02: gnu: Add s.


From: Alex Kost
Subject: Re: 02/02: gnu: Add s.
Date: Wed, 07 Jun 2017 12:29:10 +0300
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/25.2 (gnu/linux)

Mark H Weaver (2017-06-06 16:47 -0400) wrote:

> Alex Kost <address@hidden> writes:
>
>> Mark H Weaver (2017-06-04 20:15 -0400) wrote:
>>
>>>> +(define-public s
>>>> +  (let ((commit "6604341edb3a775ff94415762af3ee9bd86bfb3c")
>>>> +        (revision "1"))
>>>> +    (package
>>>> +      (name "s")
>>>> +      (version (string-append "0.0.0-" revision "." (string-take commit 
>>>> 7)))
>>>
>>> I think we should rename this package and variable name to 's-shell' or
>>> something along those lines.  's' is commonly used as a local variable
>>> name.  Single character variable names are in short supply, and I don't
>>> think we should allocate them to packages.
>>>
>>> Thoughts?
>>
>> And what about "r" package?
>
> Ah, I had forgotten about 'r'.  Thanks for reminding me :)
>
> I think we can make an exception for a package as firmly established and
> widely used as 'r'.  It's already been in Guix for a long time, and
> there are hundreds of packages based on it.
>
> However, 's' has not yet had any releases, and as with any highly
> experimental new program, chances are quite slim that it will ever gain
> a non-trivial number of users.  That's nothing personal, it's just a
> simple fact about new projects: the overwhelming majority of new
> projects never gain traction.
>
> Do we really want to permanently allocate to it one of the 25 remaining
> lowercase ASCII single-letter names?  That's prime real-estate in the
> space of possible names.  Frankly, I think it's hubris for someone to
> claim one of those names for their experimental new project.  Do we want
> to set a precedent that anyone can grab one of those single-character
> names for their pet project, regardless of whether it has any users
> besides its author?
>
>> In my opinion, package names for "r" and "s" should stay the same – I
>> think these names are expected by users.  As for the variable names,
>> they may be renamed, if it is needed.
>
> I can understand that point of view.  Recently, someone named their new
> package 'ao'.  We simply weren't able to give it that name in Guix,
> because we already have a package named 'ao' in Guix.
>
> If only 26 people in the entire world choose to give their project a
> two-letter name, then the chances are good that there will be a
> collision (c.f. birthday paradox), and one of them will need to be
> renamed.
>
> Likewise, if only 5 people in the world choose a single-letter name,
> then chances are good that there will be a collision.  Who here has the
> hubris to choose one of those names?  Do we want to enable that?

Thanks for this descriptive answer!  Now I understand your position and
I agree with it :-)

BTW, I've just recalled that there are "s" and "f" emacs packages (we
named them "emacs-s" and "emacs-f", of course).

-- 
Alex



reply via email to

[Prev in Thread] Current Thread [Next in Thread]