GNU bug report logs - #20887
'make bootstrap' now verrrry slow due to recent isearch changes

Previous Next

Full log

View this message in rfc822 format

From: Artur Malabarba <bruce.connor.am <at> gmail.com>
To: Glenn Morris <rgm <at> gnu.org>
Cc: Eli Zaretskii <eliz <at> gnu.org>, Paul Eggert <eggert <at> cs.ucla.edu>, 20887 <at> debbugs.gnu.org
Subject: bug#20887: 'make bootstrap' now verrrry slow due to recent isearch changes
Date: Wed, 24 Jun 2015 18:38:42 +0100

2015-06-24 18:15 GMT+01:00 Glenn Morris <rgm <at> gnu.org>:
> Eli Zaretskii wrote:
>
>>> Options I see
>>> 1. I could move the code to a separate file, and then only require
>>> that file as part of the isearch command, so it wouldn't be loaded at
>>> compile time.
>>> 2. I could just save that char-table directly in the file. It would be
>>> large, but it should speed up the loading by a lot.
>>
>> I think 1 is better.  Make that variable autoloaded in isearch.el, and
>> the problem should be solved.
>
> (I didn't look at the details of this case, but autoloading variables
> tends to just dump the entire definition in the target file, so that
> might not help.)

I'll actually autoload the function that does the conversion.

> Do note that
>
> time ./src/emacs -Q -batch -l ./lisp/isearch.el
>
> takes ~ 90 seconds here (versus ~ 0.1 sec for the .elc version).
>
> So it's still leaving a time-bomb around if anyone ever loads the
> uncompiled version of wherever that code ends up.

I don't know. If I run that command I get similar results to you (~ 50
sec). But now I've moved the code to its own file, and running that
command on the new file yields 3 seconds (which is roughly the same I
get from evaluating it inside Emacs).
So maybe it's not such a big time bomb.

I'll push the change and we'll see if it helps.

> (Why is it so dog slow?)

No idea! But it's something related to isearch.el.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.