Gnus cleanup

Gnus cleanup

[Lars Ingebrigtsen]

I'm starting to do the Gnus de-compat cleanup now, I think.  Removing
the compat calls is easy enough, but should I also remove the compat
functions themselves?

For instance:

  (defalias 'gnus-window-inside-pixel-edges
    (if (fboundp 'window-inside-pixel-edges)
        'window-inside-pixel-edges
      'window-pixel-edges))

In Emacs, we usually mark functions as obsolete for a few years before
removing them, but I kinda feel like it's rather unlikely that there
will be any other usages of gnus-window-inside-pixel-edges after I clean
up the Gnus sources.

So I could just remove the compat function to.  But it is theoretically
feasible that somebody out there are calling
gnus-window-inside-pixel-edges in their own code.

(gnus-window-inside-pixel-edges is just one example.  There's like a
hundred of these...)

So...  just mark the compat functions as obsolete, or nuke them?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Gnus cleanup

[Michael Albinus]

Lars Ingebrigtsen <larsi@gnus.org> writes:

Hi Lars,

> So I could just remove the compat function to.  But it is theoretically
> feasible that somebody out there are calling
> gnus-window-inside-pixel-edges in their own code.
>
> So...  just mark the compat functions as obsolete, or nuke them?

In Tramp, I have nuked all compat functions some weeks ago. One user had
to change the call of tramp-compat-split-string to split-string, that's
all what I've heard.

But maybe there's a horde of users out there, willing to beat me when I
show up ...

Best regards, Michael.

Re: Gnus cleanup

[Stefan Monnier]

> So I could just remove the compat function to.  But it is theoretically
> feasible that somebody out there are calling
> gnus-window-inside-pixel-edges in their own code.

What I've used in the past goes along the lines of:
- Use my gut-feeling to decide whether to keep the obsolete compat
  function, erring on the side of "removal" rather than on the "safe" side.
- Some of the removed functions may later be re-added based on
  bug-reports from user.

And sometimes it's easier to just remove them all or just keep them all.


        Stefan

Re: Gnus cleanup

[Lars Ingebrigtsen]

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

> What I've used in the past goes along the lines of:
> - Use my gut-feeling to decide whether to keep the obsolete compat
>   function, erring on the side of "removal" rather than on the "safe" side.
> - Some of the removed functions may later be re-added based on
>   bug-reports from user.

Ok, I'll just be nuking (most of) the compat functions, and we can
re-add if it turns out to be a problem.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Gnus cleanup

[Greg Troxel]

[-- Attachment #1: Type: text/plain, Size: 1285 bytes --]


Lars Ingebrigtsen <larsi@gnus.org> writes:

> In Emacs, we usually mark functions as obsolete for a few years before
> removing them, but I kinda feel like it's rather unlikely that there
> will be any other usages of gnus-window-inside-pixel-edges after I clean
> up the Gnus sources.

I'm someone who doesn't want to run emacs-current (because I want a
release), and I have some interest in running -current gnus, although
less so these days given how stable it is.  From my point of view, what
would be useful is retaining compat functions for the latest emacs 24
formal release, until emacs 25 is released.  I gather you don't want to
do that, but I think it would open up the ability to use current gnus to
a much wider audience for very little trouble, and that would help with
testing.

But, from where I sit, removing the compat functions from the gnus
sources within emacs-current doesn't hurt at all, once you are taking
out the calls to those functions.

It may be that there is almost nothing needed by gnus in emacs-25 not
available in emacs-24, or that there is some generic compat library to
provide the missing functions, and it will be easy enough to run emacs
24 with -current gnus.

I'll refrain from commenting on this in the future....

Sent from my Emacs 24.3.1

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 180 bytes --]

Re: Gnus cleanup

[Lars Ingebrigtsen]

I've now completed the first pass over the Gnus sources, removing almost
all XEmacs compat code.  It has, of course, not all been tested very
well, but it should be "obviously correct".  :-)  Virtually all of the
changes involve either removing the first part after
(if (featurep 'xemacs) ...) and leaving the "else" part in place.

The compat function removal can be a little bit more invasive, but it's
basically usually just replacing gnus-called-interactively-p with
called-interactively-p, so it should also be "obviously" correct.

But please do submit bug reports if this all blows up.

The next thing I was going to do is grep for "boundp" and look whether
the code looks like compat code or not.  This may introduce actual bugs,
because various builds of Emacs may have various things defined, but
I'll try to be careful.

But instead of jumping straight into that, I think I'm going to make the
Gnus build warning free first, which should help when doing cleanup, I
think.  But, again, these will be "real changes", and may introduce
bugs.  But I hope not.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Gnus cleanup

[Frank Haun]

On Fri, 12 Feb 2016 15:19:25 +1100, Lars Ingebrigtsen wrote:

> But please do submit bug reports if this all blows up.

Header coding seems broken. Something like this:

| Subject: grüne Bäume

goes total wrong.

Now when I reading those test posts, I get:

| replace-regexp-in-string: Invalid regexp: "Invalid back reference"

I've made a bug report.

Frank

Re: Gnus cleanup

[Eric Abrahamsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

[...]

> But please do submit bug reports if this all blows up.

I just pulled and remade Emacs, and when I started Gnus, all my nnimap
groups were displayed and asterisked (nntp was fine). Checking new news
just zips along and tells me everything's fine, no news.

Manually entering each group restores the correct marks to the group,
and after that they work fine. I've got an awful lot of groups, though,
and I wonder if there isn't a way to do this all in one fell swoop.

I'm only assuming this happened because of the cleanups, but I think
it's a reasonable assumption to make!

Eric

Re: Gnus cleanup

[Frank Haun]

On Fri, 12 Feb 2016 15:19:25 +1100, Lars Ingebrigtsen wrote:

> I've now completed the first pass over the Gnus sources,

First to say, thanks for that work. It seems Gnus gets faster and
faster. Does it really? I use Gnus on a Raspberry Pi 2 and every speed
improvement is very welcome. ;-)

Something I noticed is, switching back to Gnus before the cleanup isn't
a good idea. E.g. `gnus-summary-refer-parent-article' doesn't work then.

Frank

Fup2 gmane.emacs.gnus.general

Re: Gnus cleanup

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

In regard to the Raspberry Pi, it is not a good choice among
single board computers.  We grade it as "fatal flaw".
See fsf.org/resources/hw/single-board-computers.

-- 
Dr Richard Stallman
President, Free Software Foundation (gnu.org, fsf.org)
Internet Hall-of-Famer (internethalloffame.org)
Skype: No way! See stallman.org/skype.html.

Re: Gnus cleanup

[Lars Ingebrigtsen]

Frank Haun <fh@fhaun.de> writes:

> First to say, thanks for that work. It seems Gnus gets faster and
> faster. Does it really? I use Gnus on a Raspberry Pi 2 and every speed
> improvement is very welcome. ;-)

I think the speed impact of these cleanups should be pretty small.  :-)

> Something I noticed is, switching back to Gnus before the cleanup isn't
> a good idea. E.g. `gnus-summary-refer-parent-article' doesn't work then.

There should be no functional changes.  Could you be loading parts of an
old Gnus and parts of a new one?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Gnus cleanup

[Ted Zlatanov]

On Fri, 12 Feb 2016 15:19:25 +1100 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> I've now completed the first pass over the Gnus sources, removing almost
LI> all XEmacs compat code.  It has, of course, not all been tested very
LI> well, but it should be "obviously correct".  :-)  Virtually all of the
LI> changes involve either removing the first part after
LI> (if (featurep 'xemacs) ...) and leaving the "else" part in place.

Thank you for working on this. It really has made the code a friendlier
place.

Ted