Slow generation of Org manual

Slow generation of Org manual

[Eli Zaretskii]

The generation of the Org manual from org.org on the master branch is
again taking a long time.  On one system I use (which is relatively
slow), it takes a whopping 4.5 min.  On a much faster system it still
takes more than 20 sec.

Can this be sped-up, please?

Re: Slow generation of Org manual

[Ihor Radchenko]

Eli Zaretskii <eliz@gnu.org> writes:

> The generation of the Org manual from org.org on the master branch is
> again taking a long time.  On one system I use (which is relatively
> slow), it takes a whopping 4.5 min.  On a much faster system it still
> takes more than 20 sec.
>
> Can this be sped-up, please?

I do not think that it has anything to do with Org.
I also observed slow generation of the manual on master, but it was a
result of Org manual being generated _before_ Org mode is compiled. If I
do

time (touch doc/misc/org.org && make)

_after_ full compilation, I am getting

real	0m10.850s
user	0m10.548s
sys	0m0.293s

This is the usual time I saw in the earlier versions, after we bumped
the gc-cons-threshold.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: Slow generation of Org manual

[Eli Zaretskii]

> From: Ihor Radchenko <yantar92@posteo.net>
> Cc: emacs-devel@gnu.org
> Date: Sat, 29 Jun 2024 15:11:41 +0000
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > The generation of the Org manual from org.org on the master branch is
> > again taking a long time.  On one system I use (which is relatively
> > slow), it takes a whopping 4.5 min.  On a much faster system it still
> > takes more than 20 sec.
> >
> > Can this be sped-up, please?
> 
> I do not think that it has anything to do with Org.
> I also observed slow generation of the manual on master, but it was a
> result of Org manual being generated _before_ Org mode is compiled.

That's not what I see here.  Org is already byte-compiled when the Org
manual is generated.  So this cannot be the reason.

> If I do
> 
> time (touch doc/misc/org.org && make)
> 
> _after_ full compilation, I am getting
> 
> real	0m10.850s
> user	0m10.548s
> sys	0m0.293s

I'm guessing this is on a fast machine, in an optimized build, and
probably also with native compilation, right?

> This is the usual time I saw in the earlier versions, after we bumped
> the gc-cons-threshold.

Which earlier versions?  Does that include Emacs 29?

Anyway, something has clearly changed, since it never took so long,
AFAIR.  Did the Org manual in the version that is now on the emacs-30
branch change significantly wrt what we had in Emacs 29?

Re: Slow generation of Org manual

[Ihor Radchenko]

Eli Zaretskii <eliz@gnu.org> writes:

>> _after_ full compilation, I am getting
>> 
>> real	0m10.850s
>> user	0m10.548s
>> sys	0m0.293s
>
> I'm guessing this is on a fast machine, in an optimized build, and
> probably also with native compilation, right?

Sure. But the key point is that it is a similar number with what I
recall from earlier.

>> This is the usual time I saw in the earlier versions, after we bumped
>> the gc-cons-threshold.
>
> Which earlier versions?  Does that include Emacs 29?

Yes. Since around b3cbcebd9aa.

> Anyway, something has clearly changed, since it never took so long,
> AFAIR.  Did the Org manual in the version that is now on the emacs-30
> branch change significantly wrt what we had in Emacs 29?

I do not see any difference on my side.
Many things changed in the parser. Some things changed in the manual.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>