> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> | I'll let you take a look at it :)
Lars> If I have to look outside resizeCurrentBuffer, I have no clue.
So look there first. The resizeInsets thing is suspiciously outside of
all if clauses.
JMarc
> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> Yes, since a inset only (currently) exits in one buffer at a
Lars> time, the lyxtext that it contains, will have the same
Lars> "geometry" as the enclosing lyxtext, and should be usable right
Lars> away. There should be no need
On 13-Dec-2001 Lars Gullik Bjønnes wrote:
>>> bv_->text = textcache.findFit(buffer_, workarea_.workWidth());
>| Well Lars wrote that particullary part of code so, Lars?
>| But my answer would be no it is obviously not enough.
>
> Why?
Well I have to admit I didn't look at the code but do we
On 13-Dec-2001 Jean-Marc Lasgouttes wrote:
> Isn't it what
> bv_->text = textcache.findFit(buffer_, workarea_.workWidth());
> does?
Well Lars wrote that particullary part of code so, Lars?
But my answer would be no it is obviously not enough.
Jug
--
-._-._-._-._-._-._-._-._-._-._-.
> "Juergen" == Juergen Vigna <[EMAIL PROTECTED]> writes:
Juergen> Well but you don't know if it changed size while inside
Juergen> another buffer, do you? So when you're switching buffer you
Juergen> have to:
AFAIK, if you find a textcache with the right buffer and the right
width, you can u
On 13-Dec-2001 Jean-Marc Lasgouttes wrote:
> Juergen> Well maybe. In pre-InsetText times a resize was not so heavy
> Juergen> as it only had to rebreak the actual LyXText! Now if you use
> Juergen> a lot of insets it is MUCH heavier. We surely can do some
> Juergen> profiling here and make a res
> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> | Let's say we have to handle one problem at a time.
Lars> Sure, but if you do a fix that covers the real problem, the real
Lars> problem will never be fixed.
I fixed the problem that resize take a quadratic time when they sh
> "Juergen" == Juergen Vigna <[EMAIL PROTECTED]> writes:
Juergen> On 13-Dec-2001 Jean-Marc Lasgouttes wrote:
>> I did some profiling (I did a lot of document switches to give it a
>> higher weight than the initial loading phase) and a lot of time is
>> spent in Buffer::resizeInsets. I notice
On 13-Dec-2001 Jean-Marc Lasgouttes wrote:
> I did some profiling (I did a lot of document switches to give it a
> higher weight than the initial loading phase) and a lot of time is
> spent in Buffer::resizeInsets. I notice too that this method is called
> unconditionally in BufferView::Pimpl::r
> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> Juergen Vigna <[EMAIL PROTECTED]> writes:
>>> I don't know what JMarc did, but it seems to be the right patch
>>> for working with the 1.2 doc.
>>
Lars> | Well I would say we have the Hero of the day then ;)
Lars> _or_ are we
> "Juergen" == Juergen Vigna <[EMAIL PROTECTED]> writes:
Juergen> On 12-Dec-2001 Lars Gullik Bjønnes wrote:
>> Found a LyXText that fits: Buffer: 0x84d0aa0 Width: 667 TextCache:
>> resizeCurrentBuffer Buffer: 0x84d06b8 Width: 667
Juergen> Why does it a resize if the sizes are the same??? A
On 13-Dec-2001 Lars Gullik Bjønnes wrote:
> _or_ are we just covering the real problem?
No he really resolved a problem we had looking up paragraphs inside insets!
Jug
--
-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._
Dr. Jürgen VignaE-Mail: [EMAIL PR
On 13-Dec-2001 Lars Gullik Bjønnes wrote:
> Oh just read the code...
Oh may I do that! You're too good!
> (and it is possible that we do a resize to many)
Well you surely had a look so you should know ;)
Jug
--
-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._
> I don't know what JMarc did, but it seems to be the right
> patch for working with the 1.2 doc.
Well I would say we have the Hero of the day then ;)
Anyway IMO 2 secs is too much if you didn't change the width of the
BufferView. We just have to switch over pointers and redraw the visible
area
On 12-Dec-2001 Lars Gullik Bjønnes wrote:
> Found a LyXText that fits:
> Buffer: 0x84d0aa0 Width: 667
> TextCache: resizeCurrentBuffer
> Buffer: 0x84d06b8 Width: 667
Why does it a resize if the sizes are the same???
A resize has to recalculate ALL of the insettext!
A
latest cvs from today:
my 2 MegaBytes doc:
- loading and converting from 1.1.6: less than
3 mniutes -> no real change
- saving as 1.2 doc, closing and reopen takes
7 seconds -> was 17 seconds
- loading a short doc and switching from short to long takes
2 seconds -> was 7 seconds
I
Lars Gullik Bjønnes wrote:
> Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
>
> | One think that will give great gains on reformat speed is to avoid
> | computing the font of each character we look for. I do not remember
> | the specifics, but it is around 20-30% or cimplete time.
>
> first
> "Juergen" == Juergen Vigna <[EMAIL PROTECTED]> writes:
Juergen> On 12-Dec-2001 Herbert Voss wrote:
>>> Is the UserGuide.lyx large enough to show this behaviour on your
>>> box?
>> no
Juergen> Hmmm is your document longer or does it have more figures? Or
Juergen> maybe it's a multi-part-d
Juergen Vigna wrote:
> On 12-Dec-2001 Herbert Voss wrote:
>
>
>>>Is the UserGuide.lyx large enough to show this behaviour on your box?
>>>
>>no
>>
>
> Hmmm is your document longer or does it have more figures? Or maybe it's
> a multi-part-document? You'll understand that for us it's quite ha
On Wed, Dec 12, 2001 at 04:15:45PM +0100, Juergen Vigna wrote:
>
> On 12-Dec-2001 Herbert Voss wrote:
>
> >> Is the UserGuide.lyx large enough to show this behaviour on your box?
> >
> > no
>
> Hmmm is your document longer or does it have more figures? Or maybe it's
> a multi-part-document? Y
On 12-Dec-2001 Herbert Voss wrote:
>> Is the UserGuide.lyx large enough to show this behaviour on your box?
>
> no
Hmmm is your document longer or does it have more figures? Or maybe it's
a multi-part-document? You'll understand that for us it's quite hard to
fix stuff we cannot reproduce.
Lars Gullik Bjønnes wrote:
> Herbert Voss <[EMAIL PROTECTED]> writes:
>
> | Jean-Marc Lasgouttes wrote:
>
"Jean-Marc" == Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
>>>Jean-Marc> I did some profiling with herbert's book.
>>>Jean-Marc> First, I did not find any e
On 12-Dec-2001 Herbert Voss wrote:
> I reported a fact, nothing else.
Good! ;P
> I can live with any time it may take to convert which is less
> than making a cup of coffee.
Well we really hope we'll not get that bad #:O)
Jug
--
-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-.
Juergen Vigna wrote:
> On 12-Dec-2001 Herbert Voss wrote:
>
>
>>no and also don't understand why switching between large/short docs
>>in 1.1.6 is possible in less than a second and in 1.2. more than 6
>>seconds from short to long.
>>
>
> Well WE don't understand that too and have to profile
> "Herbert" == Herbert Voss <[EMAIL PROTECTED]> writes:
Herbert> no and also don't understand why switching between
Herbert> large/short docs in 1.1.6 is possible in less than a second
Herbert> and in 1.2. more than 6 seconds from short to long.
I did not check this one. I'll have a look. Is
On 12-Dec-2001 Herbert Voss wrote:
> no and also don't understand why switching between large/short docs
> in 1.1.6 is possible in less than a second and in 1.2. more than 6
> seconds from short to long.
Well WE don't understand that too and have to profile it before taking
measures. It's not t
Jean-Marc Lasgouttes wrote:
>>"Jean-Marc" == Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
>>
>
> Jean-Marc> I did some profiling with herbert's book.
>
> Jean-Marc> First, I did not find any eviidence of something related to
> Jean-Marc> converting from 1.1.6. THis is weird.
>
> H
> "Jean-Marc" == Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
Jean-Marc> I did some profiling with herbert's book.
Jean-Marc> First, I did not find any eviidence of something related to
Jean-Marc> converting from 1.1.6. THis is weird.
Herbert, could this be related to the time it takes
> "Juergen" == Juergen Vigna <[EMAIL PROTECTED]> writes:
Juergen> Well no this is NOT work in progress. I put that comment
Juergen> there because this update was commented out and it IS needed
Juergen> to put f.ex.: the tabular where it should be (centered
Juergen> insetead of moved to the ri
On 11-Dec-2001 Jean-Marc Lasgouttes wrote:
> This is the code which enventually invokes workWidth. In a long
> document with a lot of text insets, this is really a problem. I see
> several solutions to this problem.
>
> 1/ remove the code, since Juergen's comment suggests this is still
> work i
30 matches
Mail list logo