GNU ELPA

GNU ELPA

[Leo Liu]

Hello,

Any possibility to set up GNU ELPA in such a way that the package
archive is rebuilt without too much delay after commits?

Leo

Re: GNU ELPA

[Dmitry Gutov]

Leo Liu <sdl.web@gmail.com> writes:

> Any possibility to set up GNU ELPA in such a way that the package
> archive is rebuilt without too much delay after commits?

I'd also like to know the current state of affairs.

company-pkg.el was updated ~12 hours ago, yet the new version of the
package still hasn't been built.

Re: GNU ELPA

[Stefan Monnier]

>> Any possibility to set up GNU ELPA in such a way that the package
>> archive is rebuilt without too much delay after commits?

Yes.  The current cron job is a bit too heavy weight to run it often,
but hopefully someone can tweak it so that it looks at what has changed
to decide which packages to rebuild (instead of rebuilding the world
every time).

> I'd also like to know the current state of affairs.

It's described in elpa/packages/README.

> company-pkg.el was updated ~12 hours ago, yet the new version of the
> package still hasn't been built.

The cron job is run once per day, so you may have to wait up to 24h.
Also, the cron job may fail if the copyright lines aren't right.


        Stefan

Re: GNU ELPA

[Dmitry Gutov]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> I'd also like to know the current state of affairs.
>
> It's described in elpa/packages/README.

>> company-pkg.el was updated ~12 hours ago, yet the new version of the
>> package still hasn't been built.
>
> The cron job is run once per day, so you may have to wait up to 24h.
> Also, the cron job may fail if the copyright lines aren't right.

Thank you. The rebuild interval doesn't seem to be mentioned in README,
though.

Could you also advise on my current problem with company/.dir-locals.el?

I added it to the repository, because Emacs committers sometimes install
changes into elpa branch directly.

And yesterday I added there a setting for
`emacs-lisp-docstring-fill-column', to honor the existing formatting in
docstrings, and it's not safe-local by default.

Should I remove the file, the variable from it, or can we add an
exception to not add .dir-locals.el to packages?

Re: GNU ELPA

[Stefan Monnier]

>>> I'd also like to know the current state of affairs.
>> It's described in elpa/packages/README.
>>> company-pkg.el was updated ~12 hours ago, yet the new version of the
>>> package still hasn't been built.
>> The cron job is run once per day, so you may have to wait up to 24h.
>> Also, the cron job may fail if the copyright lines aren't right.
> Thank you.  The rebuild interval doesn't seem to be mentioned in README,
> though.

It says "nightly", which I thought was clear enough (tho it doesn't say
which timezone is used, so it ends up meaning little more than once per day).

> Could you also advise on my current problem with company/.dir-locals.el?

I don't understand what is the problem with it:

> I added it to the repository, because Emacs committers sometimes install
> changes into elpa branch directly.

Seems OK, yes.  Actually, we could/should even place one in
elpa/.dir-locals.el.

> And yesterday I added there a setting for
> `emacs-lisp-docstring-fill-column', to honor the existing formatting
> in docstrings, and it's not safe-local by default.

Hmm... feel free to mark it as safe-local, but of course that won't help
users of Emacs<=24.3.

> Should I remove the file, the variable from it, or can we add an
> exception to not add .dir-locals.el to packages?

It doesn't seem to warrant such measures.
After all, the user can just say "accept and consider this value as safe
from now on".


        Stefan

Re: GNU ELPA

[Dmitry Gutov]

On 23.03.2013 23:44, Stefan Monnier wrote:
>>>> I'd also like to know the current state of affairs.
>>> It's described in elpa/packages/README.
>>>> company-pkg.el was updated ~12 hours ago, yet the new version of the
>>>> package still hasn't been built.
>>> The cron job is run once per day, so you may have to wait up to 24h.
>>> Also, the cron job may fail if the copyright lines aren't right.
>> Thank you.  The rebuild interval doesn't seem to be mentioned in README,
>> though.
>
> It says "nightly", which I thought was clear enough (tho it doesn't say
> which timezone is used, so it ends up meaning little more than once per day).

Ah, sorry. I found the "nightly" bit when I first read your previous 
reply, but couldn't find it today, having opened elpa/README instead.

>> Could you also advise on my current problem with company/.dir-locals.el?
>
> I don't understand what is the problem with it:
>
>> I added it to the repository, because Emacs committers sometimes install
>> changes into elpa branch directly.
>
> Seems OK, yes.  Actually, we could/should even place one in
> elpa/.dir-locals.el.

I think so. Though that would force all packages with different 
formatting to grow a similar file.

>> And yesterday I added there a setting for
>> `emacs-lisp-docstring-fill-column', to honor the existing formatting
>> in docstrings, and it's not safe-local by default.
>
> Hmm... feel free to mark it as safe-local, but of course that won't help
> users of Emacs<=24.3.

My thoughts exactly.

>> Should I remove the file, the variable from it, or can we add an
>> exception to not add .dir-locals.el to packages?
>
> It doesn't seem to warrant such measures.
> After all, the user can just say "accept and consider this value as safe
> from now on".

I have a report that the prompt appears during the installation, in some 
circumstances: https://github.com/company-mode/company-mode/issues/4

Probably an issue with some third-party package, but if Steve Purcell 
has it, then any of the hundreds of people who base their Emacs 
configuration off of his are likely to encounter it, too.

Maybe that's also acceptable, I haven't decided yet.

Re: GNU ELPA

[Dmitry Gutov]

On 24.03.2013 0:52, Dmitry Gutov wrote:
> On 23.03.2013 23:44, Stefan Monnier wrote:
>>> Should I remove the file, the variable from it, or can we add an
>>> exception to not add .dir-locals.el to packages?
>>
>> It doesn't seem to warrant such measures.
>> After all, the user can just say "accept and consider this value as safe
>> from now on".
>
> I have a report that the prompt appears during the installation, in some
> circumstances: https://github.com/company-mode/company-mode/issues/4
>
> Probably an issue with some third-party package, but if Steve Purcell
> has it, then any of the hundreds of people who base their Emacs
> configuration off of his are likely to encounter it, too.

Okay, that was a local problem and it had been fixed. Sorry for the noise.