Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Michael Albinus]

blc@savannah.gnu.org (Basil L. Contovounesios) writes:

Hi,

> --- a/etc/NEWS
> +++ b/etc/NEWS
> @@ -1749,6 +1749,9 @@ tags to be considered as well.
>  *** New user option 'vc-git-log-switches'.
>  String or list of strings specifying switches for Git log under VC.
>
> +---
> +*** The macro 'vc-call' no longer evaluates its second argument twice.
> +

Being curious: this sounds rather like a bug fix. Is this worth to be
mentioned in etc/NEWS?

Best regards, Michael.

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Po Lu]

Michael Albinus <michael.albinus@gmx.de> writes:

> Being curious: this sounds rather like a bug fix. Is this worth to be
> mentioned in etc/NEWS?

FWIW I remember the docstring containing something along the lines of:
BEWARE!! FILE is evaluated twice!!

While it personally doesn't affect me, it seems likely for there to be
people who actually rely on this documented behaviour.  Is it really a
good idea to change it so arbitrarily?

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Basil L. Contovounesios]

Michael Albinus [2021-09-21 14:51 +0200] wrote:

> blc@savannah.gnu.org (Basil L. Contovounesios) writes:
>
>> --- a/etc/NEWS
>> +++ b/etc/NEWS
>> @@ -1749,6 +1749,9 @@ tags to be considered as well.
>>  *** New user option 'vc-git-log-switches'.
>>  String or list of strings specifying switches for Git log under VC.
>>
>> +---
>> +*** The macro 'vc-call' no longer evaluates its second argument twice.
>
> Being curious: this sounds rather like a bug fix.

Yes, that was my first thought as well.

> Is this worth to be mentioned in etc/NEWS?

Personally I think all of my commits are newsworthy ;).

Lars argued that the "buggy" behaviour was explicitly documented for a
long time, so it's safer to announce it: https://bugs.gnu.org/50690#8

I was happy to oblige to tickle my megalomania and because I'm just a
poor student who can't afford to be sued.

-- 
Basil

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Dmitry Gutov]

On 21.09.2021 16:45, Basil L. Contovounesios wrote:
> Lars argued that the "buggy" behaviour was explicitly documented for a
> long time, so it's safer to announce it:https://bugs.gnu.org/50690#8
> 
> I was happy to oblige to tickle my megalomania and because I'm just a
> poor student who can't afford to be sued.

I think the NEWS entry could be useful for someone who comes later and 
writes code which needs to be compatible with previous versions of 
Emacs: if double evaluation is a problem for them, they could at least 
find the NEWS entry, to see when it was fixed.

We don't usually do that for bug fixes, but since this was the 
documented behavior... *shrug*

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Michael Albinus]

"Basil L. Contovounesios" <contovob@tcd.ie> writes:

Hi Basil,

>> Is this worth to be mentioned in etc/NEWS?
>
> Personally I think all of my commits are newsworthy ;).

For sure!

> Lars argued that the "buggy" behaviour was explicitly documented for a
> long time, so it's safer to announce it: https://bugs.gnu.org/50690#8

I don't object; I was just curious :-)

Best regards, Michael.

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Eli Zaretskii]

> From: Dmitry Gutov <dgutov@yandex.ru>
> Date: Tue, 21 Sep 2021 16:52:43 +0300
> Cc: emacs-devel@gnu.org
> 
> On 21.09.2021 16:45, Basil L. Contovounesios wrote:
> > Lars argued that the "buggy" behaviour was explicitly documented for a
> > long time, so it's safer to announce it:https://bugs.gnu.org/50690#8
> > 
> > I was happy to oblige to tickle my megalomania and because I'm just a
> > poor student who can't afford to be sued.
> 
> I think the NEWS entry could be useful for someone who comes later and 
> writes code which needs to be compatible with previous versions of 
> Emacs: if double evaluation is a problem for them, they could at least 
> find the NEWS entry, to see when it was fixed.
> 
> We don't usually do that for bug fixes, but since this was the 
> documented behavior... *shrug*

Yes, it's borderline, so I see no harm in having that in NEWS.

Re: master 5b962a7: Avoid double argument evaluation in vc-call macro

[Basil L. Contovounesios]

Po Lu [2021-09-21 21:40 +0800] wrote:

> Michael Albinus <michael.albinus@gmx.de> writes:
>
>> Being curious: this sounds rather like a bug fix. Is this worth to be
>> mentioned in etc/NEWS?
>
> FWIW I remember the docstring containing something along the lines of:
> BEWARE!! FILE is evaluated twice!!
>
> While it personally doesn't affect me, it seems likely for there to be
> people who actually rely on this documented behaviour.  Is it really a
> good idea to change it so arbitrarily?

I think so.

No reasonable interpretation of "¡BEWARE! ¡IRREGULAR MACRO!" would
involve relying on this pitfall.  In most cases it would mean passing a
self-evaluating expression as the macro's second argument, to protect
against the irregularity.

Since it is trivial for users of vc-call to support both its current and
previous semantics, there is also little reason to worry about breaking
compatibility.

-- 
Basil