bug#29598: 26.0; doc of `load-history'

bug#29598: 26.0; doc of `load-history'

[Drew Adams]

Shouldn't the doc say something about the order of the elements of the
alist?  It's a history, but neither the Elisp manual nor the doc string
mentions whether the beginning or the end of the list is the oldest part
of the history of loads.


In GNU Emacs 26.0.90 (build 3, x86_64-w64-mingw32)
 of 2017-10-13
Repository revision: 906224eba147bdfc0514090064e8e8f53160f1d4
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --without-dbus --host=x86_64-w64-mingw32
 --without-compress-install 'CFLAGS=-O2 -static -g3''

bug#29598: 26.0; doc of `load-history'

[Eli Zaretskii]

> Date: Wed, 6 Dec 2017 21:55:22 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> 
> Shouldn't the doc say something about the order of the elements of the
> alist?  It's a history, but neither the Elisp manual nor the doc string
> mentions whether the beginning or the end of the list is the oldest part
> of the history of loads.

Since it's an alist, and every element should be there only once, why
does the order matter?

bug#29598: 26.0; doc of `load-history'

[Drew Adams]

> > Shouldn't the doc say something about the order of the elements of the
> > alist?  It's a history, but neither the Elisp manual nor the doc string
> > mentions whether the beginning or the end of the list is the oldest
> > part of the history of loads.
> 
> Since it's an alist, and every element should be there only once, why
> does the order matter?

Because it's a history?  We already tell users, by using
that name, that it is chronological.  What we don't tell
them is which chronological order is used.

If someone is looking for something, it helps to know whether
the list order is old-to-new or new-to-old.  And if someone
locates something of interest in the list it helps to know
whether it is the stuff that comes before or after it that
was loaded when it got loaded.

Is there some reason not to mention the order?  We do, after
all, bother to call it `*-history'.

bug#29598: 26.0; doc of `load-history'

[Drew Adams]

> >> > Shouldn't the doc say something about the order of the elements of the
> >> > alist?  It's a history, but neither the Elisp manual nor the doc string
> >> > mentions whether the beginning or the end of the list is the oldest
> >> > part of the history of loads.
> >>
> >> Since it's an alist, and every element should be there only once, why
> >> does the order matter?
> >
> > Because it's a history?  We already tell users, by using
> > that name, that it is chronological.  What we don't tell
> > them is which chronological order is used.
> >
> > If someone is looking for something, it helps to know whether
> > the list order is old-to-new or new-to-old.  And if someone
> > locates something of interest in the list it helps to know
> > whether it is the stuff that comes before or after it that
> > was loaded when it got loaded.
> >
> > Is there some reason not to mention the order?  We do, after
> > all, bother to call it `*-history'.
> 
> Do you have a use-case in mind here?  If we mention the order, we would
> need to follow that in the future, whereas if we don't we could
> potentially change it.  So unless there is a concrete need to document
> it, isn't it better to just avoid it?

I really don't understand this hesitation.
The use case is described above.  Users have a
chronological list.  Why would we _not_ want to
tell them which direction the list goes?  That's
not obvious.

If at some future point Emacs changes the behavior
(it never has), then we would, hopefully, document
the new order.

This is a very minor doc request, asking for a
word or two.

bug#29598: 26.0; doc of `load-history'

[Stefan Kangas]

Drew Adams <drew.adams@oracle.com> writes:

>> > Shouldn't the doc say something about the order of the elements of the
>> > alist?  It's a history, but neither the Elisp manual nor the doc string
>> > mentions whether the beginning or the end of the list is the oldest
>> > part of the history of loads.
>>
>> Since it's an alist, and every element should be there only once, why
>> does the order matter?
>
> Because it's a history?  We already tell users, by using
> that name, that it is chronological.  What we don't tell
> them is which chronological order is used.
>
> If someone is looking for something, it helps to know whether
> the list order is old-to-new or new-to-old.  And if someone
> locates something of interest in the list it helps to know
> whether it is the stuff that comes before or after it that
> was loaded when it got loaded.
>
> Is there some reason not to mention the order?  We do, after
> all, bother to call it `*-history'.

I agree that this doesn't sound useful to document the order, in
general.  It's as likely to confuse as to help, especially if there is
we can't even think of a concrete use case for this.

Also, if we document it, we sort of make a promise to our users that it
won't change.  And there doesn't seem to exist any good reason to do
that.

So I think this should better be closed as wontfix.

bug#29598: (no subject)

[Lars Ingebrigtsen]

Stefan Kangas <stefan@marxist.se> writes:

> So I think this should better be closed as wontfix.

I agree, so I'm closing it now.

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