[liblouis-liblouisxml] Re: Bert Frees: Use of Buffers
- From: "John J. Boyer" <john.boyer@xxxxxxxxxxxxxxxxx>
- To: liblouis-liblouisxml@xxxxxxxxxxxxx
- Date: Thu, 19 Jan 2012 08:07:38 -0600
Bert,
Thanks. This will be very helpful. I will have more questions. Your
explanations can serve as the start of documentation.
I'll find out about performance as I continue to implement the utd
functions.
John
On Thu, Jan 19, 2012 at 02:52:53PM +0100, Bert Frees wrote:
> Hi John,
>
> outbuf1 is the buffer that was already there (used to be just outbuf I
> think). I've added 2 additional buffers that come after the first.
> Unlike the first, the 2nd and 3th can be bypassed. They are only
> needed in some special cases, which I will explain below:
>
> Special case 1: The setting printPageNumberRange is set to yes, and
> the setting printPageNumberAt is set to top.
>
> This means that the whole range of print pages contained in a Braille
> page is displayed at the top of the page. This is used a lot in
> Belgium. For example, if a braille page contains the last part of
> print page 10 and the first part of print page 11, the first line of
> the braille page will say: 10-11.
>
> Because in this case, the first line of the page can not be created
> until the rest of the page is finished, lines 2 till N are first
> written to outbuf2. Then, the first line is written, bypassing
> outbuf2. Then, outbuf2 is read out. All this is done in the function
> nextBraillePage.
>
> Special case 2: For some element, one or more of the style attributes
> dontSplit, keepWithNext or orphanControl are set to yes.
>
> Processing such elements sometimes requires multiple iterations. For
> example, let's say dontSplit is set to yes. Before writing the
> element, the whole 'state' of the program (including outbuf1 and
> outbuf2) is stored. Then, outbuf3 is enabled and the first iteration
> is done. If the element fits on the Braille page, outbuf3 is read out
> and we can move on. If the element doesn't fit on the Braille page,
> the previous state is restored, outbuf3 is discarded and we do a
> second iteration. This time, we start the element on a new Braille
> page so that it will fit on this page. All this is done in
> transcribe_paragraph.c
>
> I still like the idea of cascading multiple buffers, but I'm not sure
> if my implementation is very good, performance-wise. You may have to
> look at that.
>
> Best,
> Bert
>
>
> On 18-jan-12, at 19:33, John J. Boyer wrote:
>
> >Hi Bert,
> >
> >I am putting your changes into the utd functions in liblouisutdml.
> >Could
> >you explain the use of the three buffers, outbuf1 outbuf2 outbuf3
> >
> >Incidentally, I have moved the buffers off the stack and into
> >user_data.
> >This makes them available for the string translation functions as well
> >for file translation functions. They also come in handy for utd and
> >for
> >converting utd to other formats.
> >
> >Thanks,
> >John
>
--
John J. Boyer; President, Chief Software Developer
Abilitiessoft, Inc.
http://www.abilitiessoft.com
Madison, Wisconsin USA
Developing software for people with disabilities
For a description of the software, to download it and links to
project pages go to http://www.abilitiessoft.com
Other related posts:
