Recent problems with `make bootstrap'.

Recent problems with `make bootstrap'.

[Luc Teirlinck]

Due to a recent change in the bootstrapping process, bootstrapping no
longer fails, but proceeds as usual, if one of the manuals fails to
build, without any noticeable warning to the user.  The Elisp manual
no longer builds, and the breaking of the build is deliberate.  It is
trivial to fix.  As a result, people are reading and checking an
outdated version of the manual and submitting bug reports and comments
about the outdated version.

Before, if a manual failed to build, it immediately got detected and
fixed in a matter of hours.  Building manuals gets broken rather
frequently through all kinds of typos or stupid inadvertent omissions.
So the Emacs manual or info.texi could start failing to build and no
longer get updated at any time without anybody noticing.

This wastes the time of people who volunteer time to check the manuals
and seriously interferes with the checking of the manuals.  This
apparent lack of concern for people's time is discouraging.

Why not revert Revision 1.299 to Makefile.in and get rid of these two
xref's pointing nowhere and quit wasting people's time?  I have had to
do that in my private version of Emacs, since otherwise checking the
manuals would be hopeless.

I do not "morally disapprove" of xref's pointing nowhere, but I can
not see how they are an important feature.  They are not "wrong", just
useless.  We _are_ going to remember about those two variables.

Sincerely,

Luc.

Re: Recent problems with `make bootstrap'.

[Han Boetes]

Luc Teirlinck wrote:
> Why not revert Revision 1.299 to Makefile.in and get rid of
> these two xref's pointing nowhere and quit wasting people's
> time? I have had to do that in my private version of Emacs,
> since otherwise checking the manuals would be hopeless.

It may be nice if the things that are most likely to break are
compiled first. Is that an idea?



# Han

Re: Recent problems with `make bootstrap'.

[Luc Teirlinck]

Han Boetes wrote:

   It may be nice if the things that are most likely to break are
   compiled first. Is that an idea?

The .texi files are compiled very early.  So if they broke
bootstrapping, you got the error very early.  They are usually trivial
to fix for somebody knowing Texinfo.  Somebody not knowing Texinfo
could post to emacs-pretest-bug or emacs-devel and it would usually
get fixed pretty fast.

Sincerely,

Luc.

Re: Recent problems with `make bootstrap'.

[Thien-Thi Nguyen]

   From: Luc Teirlinck <teirllm@dms.auburn.edu>
   Date: Wed, 8 Dec 2004 22:19:34 -0600 (CST)

   quit wasting people's time?

all time is wasted because all time cannot be conserved.  for
example, i have just wasted some time writing `line-spacing' docs.

thi

Re: Recent problems with `make bootstrap'.

[David Kastrup]

Thien-Thi Nguyen <ttn@glug.org> writes:

>    From: Luc Teirlinck <teirllm@dms.auburn.edu>
>    Date: Wed, 8 Dec 2004 22:19:34 -0600 (CST)
>
>    quit wasting people's time?
>
> all time is wasted because all time cannot be conserved.  for
> example, i have just wasted some time writing `line-spacing' docs.

Objection.  You have _invested_ some time writing `line-spacing' docs.
While you personally are not likely to get a positive return on this
particular investment (even though occasionally writing docs may
actually make one understand the issue at hand and save time later),
you are getting a positive return of your time from other people that
write other docs saving you time in other areas.

This overall gain in efficiency (which is in stark contrast to you
investing time to find out something about proprietary software which
you can't contribute back) is one of the main arguments for the "Open
Source" programmar school (yes, I know I should go wash out my mouth
with Zope).  It _is_ an important aspect in the context of free
software, though not necessarily the most important one.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: Recent problems with `make bootstrap'.

[Luc Teirlinck]

Thien-Thi Nguyen wrote:

   all time is wasted because all time cannot be conserved.  for
   example, i have just wasted some time writing `line-spacing' docs.

Which most people will not get to see until you manage to write some
documentation for `vertical-scroll-bar' as well.  Even after that,
manuals could remain stale for ages because somebody once again forgot
an `@end defvar", forgot to double up an `@' in an email address or
whatever.  Worse, more xrefs pointing nowhere could be deliberately
introduced.

In as far as `vertical-scroll-bar' is concerned, this and the related
`default-vertical-scroll-bar' where introduced by Kim Storm on
2003-05-24.  Grepping through the sources, I see all the things
necessary for _definition_.  I could not find any evidence of
implementation, although there were so many false hits that I might
have overlooked something.  The docstring for `vertical-scroll-bar' is
definitely completely wrong.

I was told that Kim was writing documentation for `line-spacing' (but
not for `vertical-scroll-bar').  I would guess that probably Kim knows
what the deal with `vertical-scroll-bar' is.

Sincerely,

Luc.

Re: Recent problems with `make bootstrap'.

[Thien-Thi Nguyen]

   From: David Kastrup <dak@gnu.org>
   Date: Thu, 09 Dec 2004 19:29:38 +0100

   Objection.  You have _invested_ some time writing `line-spacing' docs.

i passed through the same chunk of time as everyone else (sentient).
it was not my own to invest.  however, it is true i spent my own energy.

   While you personally are not likely to get a positive return on this
   particular investment (even though occasionally writing docs may
   actually make one understand the issue at hand and save time later),
   you are getting a positive return of your time from other people that
   write other docs saving you time in other areas.

i have learned that `line-spacing' the frame parameter and
`line-spacing' the permanently buffer-local variable are different
things.  i got to play a little w/ emacs in the process, experimenting.
it was fun.  perhaps my understanding of these concepts is not thorough
enough to write authoritative documentation, but i wrote what i observed
anyway.  hopefully if there are problems people will find and fix them.

thi

Re: Recent problems with `make bootstrap'.

[Thien-Thi Nguyen]

   From: Luc Teirlinck <teirllm@dms.auburn.edu>
   Date: Thu, 9 Dec 2004 12:35:49 -0600 (CST)

   Which most people will not get to see until you manage to write some
   documentation for `vertical-scroll-bar' as well.

if i had started w/ `vertical-scroll-bar' we could make the same
comment about `line-spacing'. :-)  i will see what i can do.

thi

Re: Recent problems with `make bootstrap'.

[Eli Zaretskii]

> Date: Wed, 8 Dec 2004 22:19:34 -0600 (CST)
> From: Luc Teirlinck <teirllm@dms.auburn.edu>
> 
> Why not revert Revision 1.299 to Makefile.in and get rid of these two
> xref's pointing nowhere and quit wasting people's time?  I have had to
> do that in my private version of Emacs, since otherwise checking the
> manuals would be hopeless.

I agree that ignoring build problems is not a good idea, but I think
that instead of removing invalid xref's (or even arguing about their
usefulness), we should simply write the missing docs.  Until the
missing docs are written (assuming that the 1.30 change is reverted),
you can work around the problem by saying

	make MAKEINFO='makeinfo --force'