Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
| I attach the patch instead of commiting it, because I want to have
| a virtual nod from Lars before. I would not want to break the carefully
| crafted compatibility reading...
I'd like to wait for 1.2CVS for this, exaclyt for the reason above.
> > It is not attached, if I may object.
And this time I forgot the new files...
--
Those who desire to give up Freedom in order to gain Security,
will not have, nor do they deserve, either one. (T. Jefferson)
Index: math_fontinset.C
===
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes:
Andre> On Thu, Mar 28, 2002 at 01:27:51PM +0100, Jean-Marc Lasgouttes
Andre> wrote:
>> The problem is probably to see what kind of reasonable user
>> interface we can give to this.
Andre> I attach a diff of my 'fontinset' tree vs CVS hea
On Thu, Mar 28, 2002 at 01:27:51PM +0100, Jean-Marc Lasgouttes wrote:
> The problem is probably to see what kind of reasonable user interface we
> can give to this.
I attach a diff of my 'fontinset' tree vs CVS head. Just try it.
C-m, and type \bf, \textrm etc.
> Andre> This problem, e.g. is mag
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes:
Andre> I fiddled around with "font insets" for mathed lately, and by
Andre> now I am fairly convinced that this is the Right Thing to do.
Andre> It is much more "logical Markup", it is basically what LaTeX
Andre> does, and it is what Docb
On Thu, Mar 28, 2002 at 12:21:20PM +0100, Jean-Marc Lasgouttes wrote:
> However, it seem that the huge amount of font tokens we get in this
> case is related to the fact that textinsets do not make any effort to
> reduce their fonts agains outside world (look at UserGuide.lyx...).
>
> Presumably
> "Allan" == Allan Rae <[EMAIL PROTECTED]> writes:
Allan> On Thu, 28 Mar 2002, Jean-Marc Lasgouttes wrote: [...]
>> Therefore I moved the fonts, insets and backslash (useful for ERT)
>> tests to the top of the method.
>>
>> The result is that Buffer::parseSingleLyXformat2Token takes 16.2%
>>
On Thu, 28 Mar 2002, Jean-Marc Lasgouttes wrote:
[...]
> Therefore I moved the fonts, insets and backslash (useful for ERT) tests
> to the top of the method.
>
> The result is that Buffer::parseSingleLyXformat2Token takes 16.2% of the
> time, and the 404503 string comparisons now take less time th
Here is a small experimentation I did. It is known that parsing of
very large files is slowed down by the fact that we do a whole lot
of string compares instead of a good old binary search like in lyxlex.
Typically, when loading a big file from Herbert,
Buffer::parseSingleLyXformat2Token takes 25
