bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

> > But I again tried Emacs 24.5 (release), and the problem is still
> > there - reproducible over and over.
> 
> I think you've rediscovered bug#21055 and possibly also bug#14125.
> Both were solved in Emacs 25.
> 
> So I think we can safely mark this bug done.

OK.  But can you explain (for my learning) why I saw the
problem earlier for all builds starting with Emacs 24.5
and now I see it only for Emacs 24.5?

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Mon, 25 Dec 2017 09:38:06 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 29839@debbugs.gnu.org
> 
> > > But I again tried Emacs 24.5 (release), and the problem is still
> > > there - reproducible over and over.
> > 
> > I think you've rediscovered bug#21055 and possibly also bug#14125.
> > Both were solved in Emacs 25.
> > 
> > So I think we can safely mark this bug done.
> 
> OK.  But can you explain (for my learning) why I saw the
> problem earlier for all builds starting with Emacs 24.5
> and now I see it only for Emacs 24.5?

No, I can't.  If it's the problem I think you saw, it is fixed in
Emacs 25.1 and later.

How sure are you that you indeed tried all the versions, and not just
24.5?  Or maybe you tried that with an old elisp.info?

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

> > > How sure are you that you indeed tried all the versions, and not just
> > > 24.5?  Or maybe you tried that with an old elisp.info?
> >
> > What do you mean by "all the versions"?
> 
> All the versions of Emacs later than 24.5 that you have on that
> system.
> 
> > I tried all of the versions that I mentioned, both originally
> > and this last time (after you asked to try again).  I am sure
> > of that.  And I wouldn't even know how to try "with an old
> > elisp.info".  As I said, I used `emacs -Q'.  Could that "try
> > with an old elisp.info"?
> 
> How did you make sure all the versions of Emacs you tried displayed
> the same elisp.info file?  Each Emacs version comes with its own
> elisp.info, and you might also have elisp.info files from older
> versions.

I have no idea if I "made sure" of that.  I did nothing, other
than use `emacs -Q' then `C-h i', choose elisp, and `g Sequence
Functions', then try to follow the links regarding Nth.

The info files for a given Emacs build are in the location
predefined by the build - I don't move them.  E.g., for
25.3.1 they are here: Z:\Emacs-25.3.1\share\info.  The file
`elisp.info' in that directory is no doubt what was provided
with Emacs 25.3.1.  I don't add or subtract to whatever is
in the `info' directory.
 
> With Emacs 24.5, it's expected that you will see the problem.  Your
> question was how could that happen with later versions.  I think it
> could only happen with later versions if you tried to display
> elisp.info which was produced by old versions of makeinfo.

See above.  I didn't try to display anything special.
`emacs -Q' took care of what was displayed, not I.

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Tue, 26 Dec 2017 13:37:06 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 29839@debbugs.gnu.org
> 
> > How did you make sure all the versions of Emacs you tried displayed
> > the same elisp.info file?  Each Emacs version comes with its own
> > elisp.info, and you might also have elisp.info files from older
> > versions.
> 
> I have no idea if I "made sure" of that.  I did nothing, other
> than use `emacs -Q' then `C-h i', choose elisp, and `g Sequence
> Functions', then try to follow the links regarding Nth.
> 
> The info files for a given Emacs build are in the location
> predefined by the build - I don't move them.  E.g., for
> 25.3.1 they are here: Z:\Emacs-25.3.1\share\info.  The file
> `elisp.info' in that directory is no doubt what was provided
> with Emacs 25.3.1.  I don't add or subtract to whatever is
> in the `info' directory.

Which means that each version of Emacs you tried displayed a different
elisp.info file.  And that in turn means that you need to look at each
of those elisp.info files to see whether they have DOS EOL format
and/or were produced by the problematic versions of Texinfo mentioned
in the bugs whose numbers I cited a few messages back.  If you still
want to understand what happened to you with the previous versions o0f
Emacs, that is (from my POV if it doesn't happen in the current
release branch, there's no reason to look into this any longer).

> > With Emacs 24.5, it's expected that you will see the problem.  Your
> > question was how could that happen with later versions.  I think it
> > could only happen with later versions if you tried to display
> > elisp.info which was produced by old versions of makeinfo.
> 
> See above.  I didn't try to display anything special.
> `emacs -Q' took care of what was displayed, not I.

The issue is not what you did, the issue is what actually happened.
And again, you asked the question about my being able to explain what
happened to you with the other Emacs versions.  I'm just trying to
help you understand that.

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

> > > > But I again tried Emacs 24.5 (release), and the problem is still
> > > > there - reproducible over and over.
> > >
> > > I think you've rediscovered bug#21055 and possibly also bug#14125.
> > > Both were solved in Emacs 25.
> > >
> > > So I think we can safely mark this bug done.
> >
> > OK.  But can you explain (for my learning) why I saw the
> > problem earlier for all builds starting with Emacs 24.5
> > and now I see it only for Emacs 24.5?
> 
> No, I can't.  If it's the problem I think you saw, it is fixed in
> Emacs 25.1 and later.
> 
> How sure are you that you indeed tried all the versions, and not just
> 24.5?  Or maybe you tried that with an old elisp.info?

What do you mean by "all the versions"?

I tried all of the versions that I mentioned, both originally
and this last time (after you asked to try again).  I am sure
of that.  And I wouldn't even know how to try "with an old
elisp.info".  As I said, I used `emacs -Q'.  Could that "try
with an old elisp.info"?

I still use Emacs 24.5 much of the time, and I see the problem
always with that release, both with and without an init file.

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Tue, 26 Dec 2017 08:58:49 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 29839@debbugs.gnu.org
> 
> > > OK.  But can you explain (for my learning) why I saw the
> > > problem earlier for all builds starting with Emacs 24.5
> > > and now I see it only for Emacs 24.5?
> > 
> > No, I can't.  If it's the problem I think you saw, it is fixed in
> > Emacs 25.1 and later.
> > 
> > How sure are you that you indeed tried all the versions, and not just
> > 24.5?  Or maybe you tried that with an old elisp.info?
> 
> What do you mean by "all the versions"?

All the versions of Emacs later than 24.5 that you have on that
system.

> I tried all of the versions that I mentioned, both originally
> and this last time (after you asked to try again).  I am sure
> of that.  And I wouldn't even know how to try "with an old
> elisp.info".  As I said, I used `emacs -Q'.  Could that "try
> with an old elisp.info"?

How did you make sure all the versions of Emacs you tried displayed
the same elisp.info file?  Each Emacs version comes with its own
elisp.info, and you might also have elisp.info files from older
versions.

> I still use Emacs 24.5 much of the time, and I see the problem
> always with that release, both with and without an init file.

With Emacs 24.5, it's expected that you will see the problem.  Your
question was how could that happen with later versions.  I think it
could only happen with later versions if you tried to display
elisp.info which was produced by old versions of makeinfo.

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

> > > Is it possible that the Info files you have installed have DOS CR-LF
> > > EOLs?  (If they do, you will see it in the mode line, where the EOL
> > > format indicator is shown.)  If so, they were produced incorrectly,
> > > and to fix them, you need to convert them to Unix EOL format.
> >
> > I see this in Emacs _releases_, with `emacs -Q', starting
> > with Emacs 24.5 - no such problem with Emacs 24.4 or prior.
> >
> > I also see it with Emacs 26 and 27 pretests (with `emacs -Q').
> 
> Can you please answer my question about the EOL format of these Info
> files?  Once again, if they have DOS EOL, the reason for the problem
> is crystal clear.

The mode line says this: U(Unix)%%-.
So no MS-DOS end-of-line translation, AFAIK.
Does that answer your question?

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Sun, 24 Dec 2017 11:58:25 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 29839@debbugs.gnu.org
> 
> > > > Is it possible that the Info files you have installed have DOS CR-LF
> > > > EOLs?  (If they do, you will see it in the mode line, where the EOL
> > > > format indicator is shown.)  If so, they were produced incorrectly,
> > > > and to fix them, you need to convert them to Unix EOL format.
> > >
> > > I see this in Emacs _releases_, with `emacs -Q', starting
> > > with Emacs 24.5 - no such problem with Emacs 24.4 or prior.
> > >
> > > I also see it with Emacs 26 and 27 pretests (with `emacs -Q').
> > 
> > Can you please answer my question about the EOL format of these Info
> > files?  Once again, if they have DOS EOL, the reason for the problem
> > is crystal clear.
> 
> The mode line says this: U(Unix)%%-.
> So no MS-DOS end-of-line translation, AFAIK.
> Does that answer your question?

Yes, it does, thanks.  It means the EOL format is not the problem.

I have now downloaded
https://alpha.gnu.org/gnu/emacs/pretest/windows/emacs-26.0.90-x86_64.zip,
extracted elisp.info from it, and looked at it in "emacs -Q" on my
system.  I cannot reproduce the problem you reported: those 2
cross-references lead me to the description of 'nth', as expected.

So now I'm even more puzzled by your report.  Can you please verify:

  . the Info file you are looking at is from the Emacs 26.0.90 pretest
  . the Emacs binary you use to look at it is from that pretest, as it
    is available now from alpha.gnu.org

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

[-- Attachment #1: Type: text/plain, Size: 1722 bytes --]

> I have now downloaded
> https://urldefense.proofpoint.com/v2/url?u=https-
> 3A__alpha.gnu.org_gnu_emacs_pretest_windows_emacs-2D26.0.90-2Dx86-
> 5F64.zip&d=DwIBAg&c=RoP1YumCXCgaWHvlZYR8PZh8Bv7qIrMUB65eapI_JnE&r=kI3P6lj
> Gv6CTHIKju0jqInF6AOwMCYRDQUmqX22rJ98&m=ld839yNfgceFGqhWZhUlKz_6rp3Ce5sZW3
> 2rdWxRU0o&s=JZebrbtRrchSmiKzs4mv2B_gkLchKlrMXilUfaHBa0s&e=,
> extracted elisp.info from it, and looked at it in "emacs -Q" on my
> system.  I cannot reproduce the problem you reported: those 2
> cross-references lead me to the description of 'nth', as expected.
> 
> So now I'm even more puzzled by your report.  Can you please verify:
> 
>   . the Info file you are looking at is from the Emacs 26.0.90 pretest
>   . the Emacs binary you use to look at it is from that pretest, as it
>     is available now from alpha.gnu.org

Very weird.

I downloaded that, tried it, and I do not see the problem now.

I again tried the other 26 pretest I have - again, no problem.
I again tried Emacs 27 (snapshot from 2017-11-30) - no problem.
I again tried Emacs 25.2 and 25.3.1 (releases) - no problem.

But I again tried Emacs 24.5 (release), and the problem is still
there - reproducible over and over.

Hope this makes sense to you; it doesn't make sense to me.

I don't understand why I saw something different before, with
the same builds (and `emacs -Q'), from what I'm seeing now.
And I don't understand why I still see the problem for Emacs
24.5 now.  I don't even think that I rebooted in between the
previous tests and these.  I'm not aware of anything having
changed.

I stepped through the debugger with Emacs 24.5.  Attached
are two screenshots showing, I think, the problem.  HTH.

[-- Attachment #2: throw-emacs-bug-29839-1.png --]
[-- Type: image/png, Size: 113780 bytes --]

[-- Attachment #3: throw-emacs-bug-29839-2.png --]
[-- Type: image/png, Size: 84091 bytes --]

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Mon, 25 Dec 2017 09:03:25 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 29839@debbugs.gnu.org
> 
> >   . the Info file you are looking at is from the Emacs 26.0.90 pretest
> >   . the Emacs binary you use to look at it is from that pretest, as it
> >     is available now from alpha.gnu.org
> 
> Very weird.
> 
> I downloaded that, tried it, and I do not see the problem now.
> 
> I again tried the other 26 pretest I have - again, no problem.
> I again tried Emacs 27 (snapshot from 2017-11-30) - no problem.
> I again tried Emacs 25.2 and 25.3.1 (releases) - no problem.
> 
> But I again tried Emacs 24.5 (release), and the problem is still
> there - reproducible over and over.

I think you've rediscovered bug#21055 and possibly also bug#14125.
Both were solved in Emacs 25.

So I think we can safely mark this bug done.

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

> > There are two links in node `Sequence Functions' that are labeled
> > `Definition of nth'.  The context suggests that by following them you
> > find information about `aref' and `nth'.  But they take you instead to
> > node `Building Lists', which does not mention either `aref' or `nth'
> and
> > does not seem to help with the origin context.
> 
> I cannot reproduce this: when I follow those 2 references, I land on
> the definition of 'nth'.  I tried Emacs 26.0.90 and the current
> release branch, and the Info files produced both for 26.0.90 and the
> current branch.
> 
> Is it possible that the Info files you have installed have DOS CR-LF
> EOLs?  (If they do, you will see it in the mode line, where the EOL
> format indicator is shown.)  If so, they were produced incorrectly,
> and to fix them, you need to convert them to Unix EOL format.

I see this in Emacs _releases_, with `emacs -Q', starting
with Emacs 24.5 - no such problem with Emacs 24.4 or prior.

I also see it with Emacs 26 and 27 pretests (with `emacs -Q').

> > This is a regression introduced in Emacs 24.  In Emacs 23 those links
> > take you to the description of function `nth' in node `List Elements'.
> 
> Emacs was changed to adapt to how the latest Texinfo versions produce
> Info files on MS-Windows.  Maybe your Info files were produced by an
> old version of makeinfo, and that no longer fits what Emacs expects.

No idea.  But for Emacs 24.5 and 25.3.1, they are
the released Windows binaries from GNU Emacs.  For the
pretests they are those from GNU Emacs.

For example, this release build:

In GNU Emacs 24.5.1 (i686-pc-mingw32)
 of 2015-04-11 on LEG570
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --prefix=/c/usr --host=i686-pc-mingw32'

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Sun, 24 Dec 2017 10:40:10 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> Cc: 29839@debbugs.gnu.org
> 
> > Is it possible that the Info files you have installed have DOS CR-LF
> > EOLs?  (If they do, you will see it in the mode line, where the EOL
> > format indicator is shown.)  If so, they were produced incorrectly,
> > and to fix them, you need to convert them to Unix EOL format.
> 
> I see this in Emacs _releases_, with `emacs -Q', starting
> with Emacs 24.5 - no such problem with Emacs 24.4 or prior.
> 
> I also see it with Emacs 26 and 27 pretests (with `emacs -Q').

Can you please answer my question about the EOL format of these Info
files?  Once again, if they have DOS EOL, the reason for the problem
is crystal clear.

bug#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Drew Adams]

There are two links in node `Sequence Functions' that are labeled
`Definition of nth'.  The context suggests that by following them you
find information about `aref' and `nth'.  But they take you instead to
node `Building Lists', which does not mention either `aref' or `nth' and
does not seem to help with the origin context.

The only occurrence of "Nth" in that node describes how "The Nth element
of the list is computed", where the list is what is returned by
`number-sequence'.  This not only doesn't help in this context; it
confuses.

This is a regression introduced in Emacs 24.  In Emacs 23 those links
take you to the description of function `nth' in node `List Elements'.

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#29839: 26.0; Bad xrefs in (elisp) `Sequence Functions'

[Eli Zaretskii]

> Date: Sun, 24 Dec 2017 09:30:20 -0800 (PST)
> From: Drew Adams <drew.adams@oracle.com>
> 
> There are two links in node `Sequence Functions' that are labeled
> `Definition of nth'.  The context suggests that by following them you
> find information about `aref' and `nth'.  But they take you instead to
> node `Building Lists', which does not mention either `aref' or `nth' and
> does not seem to help with the origin context.

I cannot reproduce this: when I follow those 2 references, I land on
the definition of 'nth'.  I tried Emacs 26.0.90 and the current
release branch, and the Info files produced both for 26.0.90 and the
current branch.

Is it possible that the Info files you have installed have DOS CR-LF
EOLs?  (If they do, you will see it in the mode line, where the EOL
format indicator is shown.)  If so, they were produced incorrectly,
and to fix them, you need to convert them to Unix EOL format.

> This is a regression introduced in Emacs 24.  In Emacs 23 those links
> take you to the description of function `nth' in node `List Elements'.

Emacs was changed to adapt to how the latest Texinfo versions produce
Info files on MS-Windows.  Maybe your Info files were produced by an
old version of makeinfo, and that no longer fits what Emacs expects.