Re: type of storeate for paragraph.

1999-11-06 Thread Lars Gullik Bjønnes
Duncan Simpson <[EMAIL PROTECTED]> writes: | InsertChar in in LyxParagraph looks like it is a lot less efficient, | expecially if I insert vast amounts of text close to the beginning of a long | paragraph. A split buffer is cheap in this instance (in my outline below | endtop just increases by

Re: type of storeate for paragraph.

1999-11-06 Thread Lars Gullik Bjønnes
Allan Rae <[EMAIL PROTECTED]> writes: | > For testing I made myself a test.lyx with a _HUGE_ single paragraph. | > (tens of pages). The char* version used 3! minuets to load this... the | > stl containers used less than 10 secs. The only place where I noticed | > that rope had any advantage was w

Re: type of storeate for paragraph.

1999-11-06 Thread Lars Gullik Bjønnes
Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: | Lars> Memory: the footprint of vector is a bit larger than char*, but | Lars> not by much, if think it is about 500k or something. | | What do you mean? Where does this bloat come from? Will it become | worse with bigger documents? This worries

Re: type of storeate for paragraph.

1999-11-06 Thread Lars Gullik Bjønnes
Alejandro Aguilar Sierra <[EMAIL PROTECTED]> writes: | On 5 Nov 1999, Jean-Marc Lasgouttes wrote: | | > Alejandro> It is relevant. I don't think deque gives many advantages | > Alejandro> over vector. | > | > I meant wrt our typical paragraph length, as Lars pointed out. | | Of course, I was t

Re: type of storeate for paragraph.

1999-01-03 Thread Allan Rae
On 5 Nov 1999, Lars Gullik Bjønnes wrote: > rope (the sgi::container) is the one that I had high expectations to, > but it was buggy and I have to edit the system files to make it > compile. Also it did not should a huge improvement. Rope is supposed to be good for very long strings or buffers t

Re: type of storeate for paragraph.

1999-01-03 Thread Duncan Simpson
Hmm... would it make sense to copy or otherwise arrange to have a paragraph subject to mangling in a larger buffer split into two pieces (before and after the cursor)? This is very efficient for insertations and deletions anywhere---all it takes is adjusting a pointer and nothing special appli

Re: type of storeate for paragraph.

1999-01-03 Thread Alejandro Aguilar Sierra
On 5 Nov 1999, Jean-Marc Lasgouttes wrote: > Alejandro> It is relevant. I don't think deque gives many advantages > Alejandro> over vector. > > I meant wrt our typical paragraph length, as Lars pointed out. Of course, I was talking also in that context. At this point it's rather not clear whic

Re: type of storeate for paragraph.

1999-01-03 Thread Jean-Marc Lasgouttes
> "Alejandro" == Alejandro Aguilar Sierra <[EMAIL PROTECTED]> writes: Alejandro> On 5 Nov 1999, Jean-Marc Lasgouttes wrote: >> I think using a container is good and the particular container is >> probably not relevant for our use. Alejandro> It is relevant. I don't think deque gives many adv

Re: type of storeate for paragraph.

1999-01-03 Thread Alejandro Aguilar Sierra
On 5 Nov 1999, Jean-Marc Lasgouttes wrote: > I think using a container is good and the particular container is > probably not relevant for our use. It is relevant. I don't think deque gives many advantages over vector. > Lars> Memory: the footprint of vector is a bit larger than char*, but > L

Re: type of storeate for paragraph.

1999-01-03 Thread Jean-Marc Lasgouttes
> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes: Lars> I have done some tests with the type of container used to store Lars> the paragraph contents in LyXParagraph. That's a good idea. Lars> I have after some testing settled on using vector. Compared to Lars> char* we get a lot o

type of storeate for paragraph.

1999-01-02 Thread Lars Gullik Bjønnes
I have done some tests with the type of container used to store the paragraph contents in LyXParagraph. I have tried with: char* vector deque rope I have after some testing settled on using vector. Compared to char* we get a lot o