emacs-orgmode
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Orgmode] Re: Bug: create latex png while export to HTML become very


From: Nick Dokos
Subject: Re: [Orgmode] Re: Bug: create latex png while export to HTML become very slow [6.35i]
Date: Mon, 26 Apr 2010 22:41:42 -0400

Carsten Dominik <address@hidden> wrote:

> Nick,
> 
> do you have time to do the following exercise?
> 
> Remove the entries in org-export-latex-default-packages-alist
> one by one to see which packages are the worst offenders?
> 
> That would be a great help.
> 

Carsten,

hyperref is the culprit: without it, the time for org-export-as-html
drop from about 9-9.5s to about 3.8s. I also tried a couple of combinations
(omit hyperref and inputenc, hyperref and inputenc and \tolerance) but the
difference there is in the noise. 


| Package omitted    | Time for org-export-as-html |
|--------------------+-----------------------------|
| none (before)      |                       13.49 |
| amssymb            |                        9.13 |
| latexsym           |                        9.33 |
| wasysym            |                        9.32 |
| marvosym           |                        9.32 |
| textcomp           |                        9.25 |
| t1enc              |                        9.43 |
| soul               |                        9.30 |
| wrapfig            |                        9.48 |
| float              |                        9.65 |
| longtable          |                        9.53 |
| graphicx           |                        9.45 |
| fixltx2e           |                        9.45 |
| hyperref           |                        3.84 |
| fontenc            |                        9.59 |
| inputenc           |                        9.08 |
| \tolerance         |                        9.07 |
| none (after)       |                        9.68 |
| none (after - 2nd) |                        9.77 |


The first "none" entry probably takes longer because of cold caches. The
"none" entries at the end of the table are more typical (and also give
an idea of typical variances). I repeated the no-hyperref measurement a
few times just to make sure: it always was around 3.8s. The rest of the
times are single-shot.

HTH,
Nick




reply via email to

[Prev in Thread] Current Thread [Next in Thread]