bug#26180: Remove motif toolkit support

bug#26180: Remove motif toolkit support

[Glenn Morris]

Package: emacs
Version: 25.2
Severity: wishlist

I'd like to suggest removing support for the Motif toolkit.

Motivation:
Supporting more toolkits (gtk2, gtk3, lucid, motif, none) makes it
harder to develop relevant low-level features.

The Motif toolkit build of Emacs seems to be very little used:
  - Fedora packages just a Gtk version
  - Debian packages Gtk and Lucid
  - Only ~20 mentions of "motif" on bug-gnu-emacs, ever. Most are just
  comparison testing

(I see https://wiki.gentoo.org/wiki/GNU_Emacs mentions it, but my
suspicion is that this is just an attempt at completeness when listing
alternatives to Gtk. Since it's listed first, I wonder if this has
caused more people to use it than otherwise would. All just speculation
on my part.)

(The "none" toolkit is also never used in practice, but I understand it
may be useful as a base/reference.)

bug#26180: Remove motif toolkit support

[martin rudalics]

 > I'd like to suggest removing support for the Motif toolkit.
 >
 > Motivation:
 > Supporting more toolkits (gtk2, gtk3, lucid, motif, none) makes it
 > harder to develop relevant low-level features.

I can't recall a case where supporting Motif, Lucid or "none" would
have possibly hindered anyone to fix one of the well known bugs with
GTK/GDK, Cairo and friends.  OTOH, I often use Motif, Lucid and "none"
to simply verify that an arbitrary feature just works or used to work
and it's only the GTK or Cairo builds which break it.

So I'm voting against your suggestion.

martin

bug#26180: Remove motif toolkit support

[Dmitry Gutov]

On 23.03.2017 10:01, martin rudalics wrote:
> OTOH, I often use Motif, Lucid and "none"
> to simply verify that an arbitrary feature just works or used to work
> and it's only the GTK or Cairo builds which break it.

If Motif were removed, you'd still have Lucid and "none" to test 
against, wouldn't you?

bug#26180: Remove motif toolkit support

[martin rudalics]

 > If Motif were removed, you'd still have Lucid and "none" to test
 > against, wouldn't you?

As a matter of fact, the only person affected by such a change in the
positive sense could be me: I'd have one build less to care about.
Still, it's sometimes reassuring when both Lucid and Motif handle a
strange case the same way.

martin

bug#26180: Remove motif toolkit support

[Glenn Morris]

martin rudalics wrote:

> As a matter of fact, the only person affected by such a change in the
> positive sense could be me: I'd have one build less to care about.

If you are really saying that only you would benefit from this
simplification (?), then I disagree with you.

> Still, it's sometimes reassuring when both Lucid and Motif handle a
> strange case the same way.

If this is the best (so far the only?) reason to keep Motif,
then it's pretty weak, IMO.

bug#26180: Remove motif toolkit support

[martin rudalics]

 >> As a matter of fact, the only person affected by such a change in the
 >> positive sense could be me: I'd have one build less to care about.
 >
 > If you are really saying that only you would benefit from this
 > simplification (?), then I disagree with you.

Can you tell me where Motif support is hindering progress in any other
area?

 >> Still, it's sometimes reassuring when both Lucid and Motif handle a
 >> strange case the same way.
 >
 > If this is the best (so far the only?) reason to keep Motif,
 > then it's pretty weak, IMO.

If it's our aim to reduce the number of bug reports, the best solution
would be to remove GTK support and make the Motif build the default one.
I still don't get it why you so fervidly want to remove sane and stable
branches from the Emacs tree.  Spring-cleaning?

martin

bug#26180: Remove motif toolkit support

[Glenn Morris]

4 years later, some progress in db23785.

bug#26180: Remove motif toolkit support

[Robert Pluim]

>>>>> On Sat, 13 Feb 2021 04:29:56 -0500, Glenn Morris <rgm@gnu.org> said:

    Glenn> 4 years later, some progress in db23785.

I do appear to have poked the hornets' nest there.

One other reason to remove motif: the lwlib library needs rebuilding
if you switch from lucid to motif or back, and our build system
doesnʼt know that, so I ended up having to clean manually.

Robert

bug#26180: Remove motif toolkit support

[Lars Ingebrigtsen]

Robert Pluim <rpluim@gmail.com> writes:

>>>>>> On Sat, 13 Feb 2021 04:29:56 -0500, Glenn Morris <rgm@gnu.org> said:
>
>     Glenn> 4 years later, some progress in db23785.
>
> I do appear to have poked the hornets' nest there.

Have there been any protests yet?  :-)

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

bug#26180: Remove motif toolkit support

[Robert Pluim]

>>>>> On Sun, 14 Feb 2021 14:32:21 +0100, Lars Ingebrigtsen <larsi@gnus.org> said:

    Lars> Robert Pluim <rpluim@gmail.com> writes:
    >>>>>>> On Sat, 13 Feb 2021 04:29:56 -0500, Glenn Morris <rgm@gnu.org> said:
    >> 
    Glenn> 4 years later, some progress in db23785.
    >> 
    >> I do appear to have poked the hornets' nest there.

    Lars> Have there been any protests yet?  :-)

Not virulent ones. The one argument Iʼve seen advanced that might
carry some weight is that CDE uses Motif, and forcing emacs under CDE
to use Lucid would make it look less integrated. I have no idea how
popular CDE is, though.

Robert

bug#26180: Remove motif toolkit support

[Lars Ingebrigtsen]

Robert Pluim <rpluim@gmail.com> writes:

> Not virulent ones. The one argument Iʼve seen advanced that might
> carry some weight is that CDE uses Motif, and forcing emacs under CDE
> to use Lucid would make it look less integrated. I have no idea how
> popular CDE is, though.

From the bug report analysis, it doesn't seem very popular.  And even
people under CDE have the option of using other toolkits, I believe.  So
I don't think that's a major argument against removing Motif support.

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

bug#26180: Remove motif toolkit support

[martin rudalics]

 > One other reason to remove motif: the lwlib library needs rebuilding
 > if you switch from lucid to motif or back,

... not if you build out of tree ...

 > and our build system
 > doesnʼt know that, so I ended up having to clean manually.

martin

bug#26180: Remove motif toolkit support

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Glenn Morris <rgm@gnu.org> writes:

> Package: emacs
> Version: 25.2
> Severity: wishlist

We agreed to keep Motif support in a discussion last December, so I'm
closing this bug.

I agree with Martin in that there isn't a single concrete case where
Motif interferes with the development of a feature.  Most of the
important code is shared with the Lucid/Athena build, since both
toolkits use Xt, and the toolkit that causes us difficulties is
typically GTK 3.  For instance, it's the only toolkit that has to be
special-cased for frame resize synchronization, since it doesn't behave
well when it doesn't know about the basic frame counter, while the
development of that feature went smoothly on every other build we
currently support.