Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Stefan Monnier]

>     * doc/lispref/processes.texi (Filter Functions): Use
>     insert-before-markers in the "ordinary" filter example, like
>     internal-default-process-filter does.

In my experience, insert-before-markers is a source of problems down the
road, so I'd recommend not to encourage its use by showing it in example
code where it's not strictly needed.


        Stefan

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Noam Postavsky]

On Tue, 30 Apr 2019 at 22:54, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>
> >     * doc/lispref/processes.texi (Filter Functions): Use
> >     insert-before-markers in the "ordinary" filter example, like
> >     internal-default-process-filter does.
>
> In my experience, insert-before-markers is a source of problems down the
> road, so I'd recommend not to encourage its use by showing it in example
> code where it's not strictly needed.

But it's an example about "mimicking the actions of the default
filter" which does currently use insert-before-markers.

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Stefan Monnier]

>> >     * doc/lispref/processes.texi (Filter Functions): Use
>> >     insert-before-markers in the "ordinary" filter example, like
>> >     internal-default-process-filter does.
>>
>> In my experience, insert-before-markers is a source of problems down the
>> road, so I'd recommend not to encourage its use by showing it in example
>> code where it's not strictly needed.
>
> But it's an example about "mimicking the actions of the default
> filter" which does currently use insert-before-markers.

But do we really want to mimic that undesirable part of the behavior?


        Stefan

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Noam Postavsky]

On Fri, 3 May 2019 at 18:50, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>
> >> >     * doc/lispref/processes.texi (Filter Functions): Use
> >> >     insert-before-markers in the "ordinary" filter example, like
> >> >     internal-default-process-filter does.
> >>
> >> In my experience, insert-before-markers is a source of problems down the
> >> road, so I'd recommend not to encourage its use by showing it in example
> >> code where it's not strictly needed.
> >
> > But it's an example about "mimicking the actions of the default
> > filter" which does currently use insert-before-markers.
>
> But do we really want to mimic that undesirable part of the behavior?

If it's undesirable, then we should stop doing it in the default
filter. But since we are currently doing it in the default filter,
what kind of problems is that leading to (I'm not aware of any)?

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Stefan Monnier]

> If it's undesirable, then we should stop doing it in the default filter.

I don't think it's worth changing long-standing behavior, no.
But showing it as an example that people may then follow is
another issue.

> But since we are currently doing it in the default filter,
> what kind of problems is that leading to (I'm not aware of any)?

The problem is that it's blunt because it applies equally to all
markers/overlays.  It's almost always the case that it's much better to
use `insert` and rely on your markers's insertion-type to control
the behavior.


        Stefan

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Noam Postavsky]

On Sat, 4 May 2019 at 09:22, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>
> > If it's undesirable, then we should stop doing it in the default filter.
>
> I don't think it's worth changing long-standing behavior, no.
> But showing it as an example that people may then follow is
> another issue.
>
> > But since we are currently doing it in the default filter,
> > what kind of problems is that leading to (I'm not aware of any)?
>
> The problem is that it's blunt because it applies equally to all
> markers/overlays.  It's almost always the case that it's much better to
> use `insert` and rely on your markers's insertion-type to control
> the behavior.

I just don't see how we can say it's almost always much better to use
`insert', and then use `insert-before-markers' by default.
"Do as we say, not as we do".

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Stefan Monnier]

> I just don't see how we can say it's almost always much better to use
> `insert', and then use `insert-before-markers' by default.

Very easy: historical accident (and the lack of
window-point-insertion-type before Emacs-23).

> "Do as we say, not as we do".

More like: "don't do the same mistake we did".


        Stefan

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Noam Postavsky]

On Tue, 7 May 2019 at 08:43, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>
> > I just don't see how we can say it's almost always much better to use
> > `insert', and then use `insert-before-markers' by default.
>
> Very easy: historical accident (and the lack of
> window-point-insertion-type before Emacs-23).
>
> > "Do as we say, not as we do".
>
> More like: "don't do the same mistake we did".

I guess I can live with that, though I kind of still have the urge to
change it and see what breaks...

Anyone else have opinions on this?

Re: [Emacs-diffs] emacs-26 ec02c73: Update process filter example (Bug#35044)

[Stefan Monnier]

> I guess I can live with that, though I kind of still have the urge to
> change it and see what breaks...

I'm all for it,


        Stefan