Re: Preview: request for testers
On Friday 12 July 2002 10:31 am, David Kastrup wrote: > > I remember in your introduction you mentioned that you'd like people > > to pay you for your LaTeX services... ? > > Sure thing, I am suffocating in all the money sent to me from LyX > users, and you are right to scald me for all the greed I have been > subjecting you to. > > I am currently living off savings and working on TeX stuff so that I > may eventually be able to continue working on it and just sustain > myself (more would be too optimistic). Instead of just looking now > for a well-paying job, for which I have offers. There is no sane > reason, except that I like working on TeX better than on automotive > driver assistence systems. > > So you work on LyX for fun for fun. Your future does not depend on > the future of LyX. I work on TeX for fun for earnest. If you think > that this entitles you to call my motivation for working with you on > LyX greed, there is little that I can say in reply. > > Shrug. I never mentioned "greed", nor do I think it. I think, good on you for giving it a go. A
Re: Preview: request for testers
Angus Leeming <[EMAIL PROTECTED]> writes: > On Friday 12 July 2002 9:41 am, David Kastrup wrote: > > > David, I work on LyX for fun. > > > > I did not particularly address you. Hmmm, what was the reason I > > was working on LyX (if you would call it that)? > > I remember in your introduction you mentioned that you'd like people > to pay you for your LaTeX services... ? Sure thing, I am suffocating in all the money sent to me from LyX users, and you are right to scald me for all the greed I have been subjecting you to. I am currently living off savings and working on TeX stuff so that I may eventually be able to continue working on it and just sustain myself (more would be too optimistic). Instead of just looking now for a well-paying job, for which I have offers. There is no sane reason, except that I like working on TeX better than on automotive driver assistence systems. So you work on LyX for fun for fun. Your future does not depend on the future of LyX. I work on TeX for fun for earnest. If you think that this entitles you to call my motivation for working with you on LyX greed, there is little that I can say in reply. Shrug. -- David Kastrup, Kriemhildstr. 15, 44793 Bochum Email: [EMAIL PROTECTED]
Re: Preview: request for testers
Angus Leeming <[EMAIL PROTECTED]> writes: > On Friday 12 July 2002 9:41 am, David Kastrup wrote: > > Well, PNG image support will mean previews will be much more fun, and > > their loading will not spend an infinite amount of time... > > Incidentally, whilst total loading time will remain high, actual > impact on the user will be negligible once I finish this "smart > loader" stuff. The previews in PPM format will take a fraction of a > second longer to appear than the ideal case of native PNG loading > once they've been generated. "Once they've been generated". Writing dozens of MBs of data is even slower than reading them. And of course, the buffer cache gets polluted with image data instead of keeping more important stuff. Anyway, GhostScript takes up quite a bit of time, too. So while you are fiddling around with the loader, don't make decisions that would make it awkward to eventually run GhostScript in daemon mode, telling it what image it should be converting next. In preview-latex, by far the slowest stage is GhostScript, and yet it is the one least impacting responsiveness, because it can cater to the requests in the needed order, depending on what is currently on-screen. dvipng offers similar functionality. -- David Kastrup, Kriemhildstr. 15, 44793 Bochum Email: [EMAIL PROTECTED]
Re: Preview: request for testers
On Friday 12 July 2002 9:41 am, David Kastrup wrote: > Well, PNG image support will mean previews will be much more fun, and > their loading will not spend an infinite amount of time... Incidentally, whilst total loading time will remain high, actual impact on the user will be negligible once I finish this "smart loader" stuff. The previews in PPM format will take a fraction of a second longer to appear than the ideal case of native PNG loading once they've been generated. Since we're only going to be loading 5 or 6 previews at a time, that's hardly important. Angus
Re: Preview: request for testers
On Friday 12 July 2002 9:41 am, David Kastrup wrote: > > David, I work on LyX for fun. > > I did not particularly address you. Hmmm, what was the reason I was > working on LyX (if you would call it that)? I remember in your introduction you mentioned that you'd like people to pay you for your LaTeX services... ? > > On a more positive note, John Levon has an almost working Qt > > frontend. That will "just work" with PNG. > > Hmmm. Image loading will depend on the toolkit used? Why would we re-invent the wheel and not use what is available to us in the toolkit? Of course image loading is done by the toolkit. The intellectual challenge is in writing a core that is GUI-independent yet doesn't dumn things down to the abilities of the stupidest toolkit. Once you've got your head around it, it's actually a pleasure coding in this way. Look at mathed. Beautiful, clean code. Doesn't know what the hell it is writing to. That must be elegant ;-) Angus
Re: Preview: request for testers
Angus Leeming <[EMAIL PROTECTED]> writes: > On Friday 12 July 2002 9:12 am, David Kastrup wrote: > > Herbert Voss <[EMAIL PROTECTED]> writes: > > > Angus Leeming wrote: > > > >>Statistics: > > > >>285 numbered equations > > > >>732 pages in the metrics file > > > >>57 MBytes for the ppm-files > > > > > > > > Experiment with gs -sDEVICE=pnmraw. Does that reduce disk footprint > > > > much? > > > > > > yes, up 14.5 MBytes > > > > This is still ridiculous. Get this PNG support going. Of course, > > this also explains why previews got slower when color support got > > added: I/O time. ppm files take up three times more space than pgm, > > and when you leave off antialiasing, you get pbm, at another factor > > of 8 in file size. I/O time for 57MBytes is far from trivial. > > > > Get this PNG support going. > > David, I work on LyX for fun. I did not particularly address you. Hmmm, what was the reason I was working on LyX (if you would call it that)? > I do not have an infiinite amount of time to spend on it. Well, PNG image support will mean previews will be much more fun, and their loading will not spend an infinite amount of time... > On a more positive note, John Levon has an almost working Qt > frontend. That will "just work" with PNG. Hmmm. Image loading will depend on the toolkit used? -- David Kastrup, Kriemhildstr. 15, 44793 Bochum Email: [EMAIL PROTECTED]
Re: Preview: request for testers
On Friday 12 July 2002 9:12 am, David Kastrup wrote: > Herbert Voss <[EMAIL PROTECTED]> writes: > > Angus Leeming wrote: > > >>Statistics: > > >>285 numbered equations > > >>732 pages in the metrics file > > >>57 MBytes for the ppm-files > > > > > > Experiment with gs -sDEVICE=pnmraw. Does that reduce disk footprint > > > much? > > > > yes, up 14.5 MBytes > > This is still ridiculous. Get this PNG support going. Of course, > this also explains why previews got slower when color support got > added: I/O time. ppm files take up three times more space than pgm, > and when you leave off antialiasing, you get pbm, at another factor > of 8 in file size. I/O time for 57MBytes is far from trivial. > > Get this PNG support going. David, I work on LyX for fun. I do not have an infiinite amount of time to spend on it. Given the options of writing something reasonable for LyX or learning how to write a new loader for an external library, guess which one I choose? On a more positive note, John Levon has an almost working Qt frontend. That will "just work" with PNG. Angus
Re: Preview: request for testers
Herbert Voss <[EMAIL PROTECTED]> writes: > Angus Leeming wrote: > > > >>Statistics: > >>285 numbered equations > >>732 pages in the metrics file > >>57 MBytes for the ppm-files > >> > > Experiment with gs -sDEVICE=pnmraw. Does that reduce disk footprint > > much? > > > yes, up 14.5 MBytes This is still ridiculous. Get this PNG support going. Of course, this also explains why previews got slower when color support got added: I/O time. ppm files take up three times more space than pgm, and when you leave off antialiasing, you get pbm, at another factor of 8 in file size. I/O time for 57MBytes is far from trivial. Get this PNG support going. -- David Kastrup, Kriemhildstr. 15, 44793 Bochum Email: [EMAIL PROTECTED]
Re: Preview: request for testers
On Friday 12 July 2002 9:06 am, Juergen Vigna wrote: > Angus Leeming wrote: > >>- equations in floats (caption) are not previewed > >>maybe all insets??? > > > > It's possible that my isInsetVisible function won't work on these. Same > > problem as the tabulars and something that Jürgen might be able to help > > with. > > Well I think you have to inform the InsetText that some update happend > and that it has to redraw itself on the row where the insets reside. > I'll try to debug this as soon as your patch is in the main line and > I can do an update. I think this shouldn't hamper the inclusion if it > works in normal text as this surely can be fixed quite fast. Well, another more mundane question. Where should I put it? bufferview_funcs.[Ch] ? Probably.
Re: Preview: request for testers
Angus Leeming wrote: >>- equations in floats (caption) are not previewed >>maybe all insets??? > > > It's possible that my isInsetVisible function won't work on these. Same > problem as the tabulars and something that Jürgen might be able to help with. Well I think you have to inform the InsetText that some update happend and that it has to redraw itself on the row where the insets reside. I'll try to debug this as soon as your patch is in the main line and I can do an update. I think this shouldn't hamper the inclusion if it works in normal text as this surely can be fixed quite fast. Jug -- -._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._ Dr. Jürgen VignaE-Mail: [EMAIL PROTECTED] Mitterstrich 151/A Tel/Fax: +39-0471-450260 / +39-0471-450253 I-39050 SteineggWeb: http://www.lyx.org/~jug -._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._
Re: Preview: request for testers
On Friday 12 July 2002 7:26 am, Andre Poenitz wrote:
> On Thu, Jul 11, 2002 at 07:04:03PM +0100, Angus Leeming wrote:
> > // top_y is not const because it's reset by getRowNearY.
> > int top_y = bv.text->first_y;
> > Row const * row = bv.text->getRowNearY(top_y);
> >
> > int const bv_height = bv.painter().paperHeight();
> > int height = row->height();
> > rows.push_back(row);
> >
> > while (height < bv_height) {
> > row = row->next();
> > + if (!row)
> > + break;
> > height += row->height();
> > rows.push_back(row);
> > }
> >
> > return rows;
> > }
>
> What about writing this as
>
> for (int height = 0; row && height < bv_height; row = row->next()) {
> height += row->height();
> rows.push_back(row);
> }
>
> Andre'
Cheers André. I'lll think about it ;-)
Re: Preview: request for testers
Herbert Voss wrote: > Angus Leeming wrote: > > >>> Statistics: >>> 285 numbered equations >>> 732 pages in the metrics file >>> 57 MBytes for the ppm-files >>> >> >> Experiment with gs -sDEVICE=pnmraw. Does that reduce disk footprint much? > > > > yes, up 14.5 MBytes aehm ... downto ... :-) Herbert -- http://www.lyx.org/help/
Re: Preview: request for testers
Angus Leeming wrote: >>Statistics: >>285 numbered equations >>732 pages in the metrics file >>57 MBytes for the ppm-files >> > > Experiment with gs -sDEVICE=pnmraw. Does that reduce disk footprint much? yes, up 14.5 MBytes Herbert -- http://www.lyx.org/help/
Re: Preview: request for testers
-BEGIN PGP SIGNED MESSAGE- On Friday 12 July 2002 09:00, Angus Leeming wrote: > On Thursday 11 July 2002 7:43 pm, Kornel Benko wrote: > > On Thursday 11 July 2002 20:21, Angus Leeming wrote: > > > Ok. Thanks. I'll have a look tomorrow. > > > > One more thing: my labels are now redefined, so that the referencies no > > longer match. (You know, "underlines" in labels, see attached) > > Some strange interation with underline.sty? This isn't preview-specific > however. File a bug report and badger Herbert ;-) Well, I don't think so. "underline.sty" may have to do something with display after the latex-call, but here it is the display of an internal table in lyx. (The lyx-file itself is not corrupted, lyx-1.2 is working ok) Kornel - -- Kornel Benko [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: PGP 6.5.8 iQCVAwUBPS6G/rewfbDGmeqhAQHPOAP9GZluTPcuuOubc4OeCMXPzYj93vkq9EMS nxeGUxVaqCmPH/YjiedcQsitCOSSc7n+RIFcEMzjsJRk8Uw8pvZj2RRs5I0CH17x vpRwSAP73hb6q4yWQSO7z/tI9iaoxLw9yO9ipuDoOPs9fiVkNamNYT1nNuBrFyzR POZa/VSeIlg= =55vw -END PGP SIGNATURE-
Re: Preview: request for testers
Angus Leeming wrote: > On Thursday 11 July 2002 9:04 pm, Herbert Voss wrote: > >>Another small problem: >> >> >>attached are the mathbox and the preview, >>it's still a bit too deep. >> >>Herbert >> > > You mean "Original" doesn't line up with "fürht"? Then modify your copy of > lyxpreview2ppm.sh. That's why it's a shell script after all. You should need > to just set REMOVE_WS=1. sure, but this looks a bit ugly ... the preview stuff is very nice so that it should be 100% to the output which I also attached. Herbert -- http://www.lyx.org/help/
Re: Preview: request for testers
On Thursday 11 July 2002 8:57 pm, Herbert Voss wrote: > Angus Leeming wrote: > > Attached is a "smart loader" for the previews. I'd really, really like > > interested users --- that's the mathematicians out there who use > > /lots/ of > > > equations to test it out. > > > > It's meant to enable you to generate and display the previews with no > > discernible effect on LyX's responsiveness. > > nice work :-) > > I loaded one of my books with a lot of mathstuff. Only some > problems: > - it starts only when I walk with the cursor "through" some > mathboxes Yes. For some reason the first page doesn't think it's visible until the entire document is loaded. Or something. If I can't track this down properly (quite possible) the work around is to load insets for the first page anyway. Should be easy enough. > - in tabulars with matstuff in every cell, I have some > in preview and some not, not so good > sed: Couldn't open file bbox.sed > extracting metrics failed. > is the message, when I walk from cell to cell > - now I see, that tabulars are totally ignored??? Really? Small example doument please. > - stere is still the problem with the not ignored > geometry settings -> eqno is unvisible You're asking the wrong man. Ask a LaTeX expert ;-) > - the eqno is f.ex: 0.21 but it's the third chapter > and it should be 3.5, a minor problem ... Alternatively, you could hack preview.sty so that it outputs (eq:label) rather than (0.21) [eq:label David said he'd be willing to accept a patch... > - equations in floats (caption) are not previewed > maybe all insets??? It's possible that my isInsetVisible function won't work on these. Same problem as the tabulars and something that Jürgen might be able to help with. > - scrolling quickly to the end -> peng Fixed. > - restart and imidiately scrolling to the end was ok, > but now only the formulas through which I "walked" > were previewed, no the one before or behind I don't understand what you mean here. > Statistics: > 285 numbered equations > 732 pages in the metrics file > 57 MBytes for the ppm-files Experiment with gs -sDEVICE=pnmraw. Does that reduce disk footprint much? Maybe now is the time to write/use lyxpreview2png.sh. On loading, the xforms loader will have to run a conversion to ppm step first, so loading will take a little longer but not much. Disk footprint would then be O(1MB). Moreover the Qt loader will just load ;-) Angus
Re: Preview: request for testers
On Thursday 11 July 2002 9:04 pm, Herbert Voss wrote: > Another small problem: > > > attached are the mathbox and the preview, > it's still a bit too deep. > > Herbert You mean "Original" doesn't line up with "fürht"? Then modify your copy of lyxpreview2ppm.sh. That's why it's a shell script after all. You should need to just set REMOVE_WS=1. Angus
Re: Preview: request for testers
On Thursday 11 July 2002 7:43 pm, Kornel Benko wrote: > On Thursday 11 July 2002 20:21, Angus Leeming wrote: > > Ok. Thanks. I'll have a look tomorrow. > > One more thing: my labels are now redefined, so that the referencies no > longer match. (You know, "underlines" in labels, see attached) Some strange interation with underline.sty? This isn't preview-specific however. File a bug report and badger Herbert ;-) Angus
Re: Preview: request for testers
Angus Leeming wrote: > One other point (for Jürgen?): I think that a (working) isInsetVisible could > be used to advantage by BufferView::updateInset. Seems intelligent to only > redraw insets that are visible... I think you didn't understand the Row::pos() variable/function. The pos returning from a row is the _first_ position of the first character of this row inside the Row::par(). You understand that in the visible rows you may have more than one paragraph and so the use of the start/end pos cannot work as you use it because most probably the lastrow is inside a different paragraph then the firstrow. Capito ;) Jug -- -._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._ Dr. Jürgen VignaE-Mail: [EMAIL PROTECTED] Mitterstrich 151/A Tel/Fax: +39-0471-450260 / +39-0471-450253 I-39050 SteineggWeb: http://www.lyx.org/~jug -._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._-._
Re: Preview: request for testers
On Thu, Jul 11, 2002 at 07:04:03PM +0100, Angus Leeming wrote:
> // top_y is not const because it's reset by getRowNearY.
> int top_y = bv.text->first_y;
> Row const * row = bv.text->getRowNearY(top_y);
>
> int const bv_height = bv.painter().paperHeight();
> int height = row->height();
> rows.push_back(row);
>
> while (height < bv_height) {
> row = row->next();
> + if (!row)
> + break;
> height += row->height();
> rows.push_back(row);
> }
>
> return rows;
> }
What about writing this as
for (int height = 0; row && height < bv_height; row = row->next()) {
height += row->height();
rows.push_back(row);
}
Andre'
--
Those who desire to give up Freedom in order to gain Security,
will not have, nor do they deserve, either one. (T. Jefferson)
Re: Preview: request for testers
Another small problem: attached are the mathbox and the preview, it's still a bit too deep. Herbert -- http://www.lyx.org/help/
Re: Preview: request for testers
Angus Leeming wrote: > Attached is a "smart loader" for the previews. I'd really, really like > interested users --- that's the mathematicians out there who use /lots/ of > equations to test it out. > > It's meant to enable you to generate and display the previews with no > discernible effect on LyX's responsiveness. nice work :-) I loaded one of my books with a lot of mathstuff. Only some problems: - it starts only when I walk with the cursor "through" some mathboxes - in tabulars with matstuff in every cell, I have some in preview and some not, not so good sed: Couldn't open file bbox.sed extracting metrics failed. is the message, when I walk from cell to cell - now I see, that tabulars are totally ignored??? - stere is still the problem with the not ignored geometry settings -> eqno is unvisible - the eqno is f.ex: 0.21 but it's the third chapter and it should be 3.5, a minor problem ... - equations in floats (caption) are not previewed maybe all insets??? - scrolling quickly to the end -> peng - restart and imidiately scrolling to the end was ok, but now only the formulas through which I "walked" were previewed, no the one before or behind Statistics: 285 numbered equations 732 pages in the metrics file 57 MBytes for the ppm-files Herbert -- http://www.lyx.org/help/
Re: Preview: request for testers
-BEGIN PGP SIGNED MESSAGE- On Thursday 11 July 2002 20:21, Angus Leeming wrote: > Ok. Thanks. I'll have a look tomorrow. One more thing: my labels are now redefined, so that the referencies no longer match. (You know, "underlines" in labels, see attached) > Have a nice evening. You too. Kornel - -- Kornel Benko [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: PGP 6.5.8 iQCVAwUBPS3R67ewfbDGmeqhAQHkrAP/U6G4Fo3WoB8wXnHDzKXJkQ1Vxj46O5I0 fs9RF4UfYdBRzjQ20HDTP0TLFZbEQoQFGLaEVEwbXzWytQpl8doNlHGJdD7Ql57S IURANxtpukgT2OgPh2JRRaISF+AvEQdehFhRQNk59W5qC2o7MFQVpG35xTYn/+eG G04ilEM4t5U= =AbNL -END PGP SIGNATURE- test.gif Description: GIF image
Re: Preview: request for testers
On Thursday 11 July 2002 7:32 pm, Kornel Benko wrote: > On Thursday 11 July 2002 20:04, Angus Leeming wrote: > > Does this cure the problem? > > It does. At least I seem to be unable to reproduce it. > > There is one feature, which may be changed. If I load a document, > there is no update of the visible formulae on the first visible page. I > have to scroll to get this new behaviour. Ok. Thanks. I'll have a look tomorrow. Have a nice evening. A
Re: Preview: request for testers
-BEGIN PGP SIGNED MESSAGE- On Thursday 11 July 2002 20:04, Angus Leeming wrote: > Does this cure the problem? It does. At least I seem to be unable to reproduce it. There is one feature, which may be changed. If I load a document, there is no update of the visible formulae on the first visible page. I have to scroll to get this new behaviour. Nice work. Kornel - -- Kornel Benko [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: PGP 6.5.8 iQCVAwUBPS3PRLewfbDGmeqhAQEdtQP/cKnjtfv6+9e2en5+Lh/t8mAmAjf1+WVQ YOrECT4yNCyuwzVbMYszZz4kxa0WsgJzIOfzzBczihEoLhL/iNjrBjjH8GMDw2k5 u+t8KKMgsEhD0FmE1upjoGmz/hkT4jhaom70CYyVPJz0ZAjiwZ93icWbcZpMFXwz woJ8TdzayYc= =ugYj -END PGP SIGNATURE-
Re: Preview: request for testers
On Thursday 11 July 2002 7:01 pm, Kornel Benko wrote:
> On Thursday 11 July 2002 18:38, Angus Leeming wrote:
> > Attached is a "smart loader" for the previews. I'd really, really like
> > interested users --- that's the mathematicians out there who use /lots/
> > of equations to test it out.
>
> Super!
Great!
> I tried it and I like it. But on the last page I had a core dump.
Hmmm!
> gdb backtrace:
> (gdb) bt
> #0 0x402b9ab1 in kill () from /lib/libc.so.6
> #1 0x402b98a8 in raise () from /lib/libc.so.6
> #2 0x402baedb in abort () from /lib/libc.so.6
> #3 0x083f6bbb in lyx::abort ()
> #4 0x081167f8 in error_handler ()
> #5
> #6 0x083f408c in getVisibleRows ()
That means I'm doing something illegal.
> It does not look very hellpful, does it?
Does this cure the problem?
RowList const getVisibleRows(BufferView const & bv)
{
RowList rows;
// top_y is not const because it's reset by getRowNearY.
int top_y = bv.text->first_y;
Row const * row = bv.text->getRowNearY(top_y);
int const bv_height = bv.painter().paperHeight();
int height = row->height();
rows.push_back(row);
while (height < bv_height) {
row = row->next();
+ if (!row)
+ break;
height += row->height();
rows.push_back(row);
}
return rows;
}
Angus
Re: Preview: request for testers
-BEGIN PGP SIGNED MESSAGE- On Thursday 11 July 2002 18:38, Angus Leeming wrote: > Attached is a "smart loader" for the previews. I'd really, really like > interested users --- that's the mathematicians out there who use /lots/ of > equations to test it out. Super! I tried it and I like it. But on the last page I had a core dump. scrollDocView of 30a2 screen: drawFromTo 0-2b6 expose 269x2b6+0+0 lyx: SIGSEGV signal caught Sorry, you have found a bug in LyX. Please read the bug-reporting instructions in Help->Introduction and send us a bug report, if necessary. Thanks ! Bye. ... gdb backtrace: (gdb) bt #0 0x402b9ab1 in kill () from /lib/libc.so.6 #1 0x402b98a8 in raise () from /lib/libc.so.6 #2 0x402baedb in abort () from /lib/libc.so.6 #3 0x083f6bbb in lyx::abort () #4 0x081167f8 in error_handler () #5 #6 0x083f408c in getVisibleRows () #7 0x083eb29e in grfx::PreviewImage::Impl::startLoading () #8 0x084c2614 in boost::detail::function::void_function_obj_invoker0, boost::_bi::list1 > >, void>::invoke () #9 0x0841f91b in boost::signal0, int, less, boost::function0 > >::operator() () #10 0x082b870b in Timeout::emit () #11 0x082d4ec2 in C_intern_timeout_cb () #12 0x400d49c0 in fl_handle_timeouts () from /usr/X11R6/lib/libforms.so.0.89 #13 0x4008e941 in get_next_event () from /usr/X11R6/lib/libforms.so.0.89 #14 0x4008d573 in do_interaction_step () from /usr/X11R6/lib/libforms.so.0.89 #15 0x4008df79 in fl_treat_interaction_events () from /usr/X11R6/lib/libforms.so.0.89 #16 0x4008dfc6 in fl_check_forms () from /usr/X11R6/lib/libforms.so.0.89 #17 0x082e8beb in lyx_gui::start () #18 0x08116055 in LyX::LyX () #19 0x0817c70c in main () #20 0x402a89ed in __libc_start_main () from /lib/libc.so.6 It does not look very hellpful, does it? There was no editing, only scrols down. Kornel - -- Kornel Benko [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: PGP 6.5.8 iQCVAwUBPS3IAbewfbDGmeqhAQHI+AQAiSaXZf7JgnVsalsciwunt3+fwz5biKyG EOaL+LMYXnS6r0JjkcdQ8t0N7z2CoFgFllOb/Oe9cIfj+CGBT3nJl3RlGlymTieq p/C/r88dJi78A4YHtL46UvKvKGTCgEARnNY+du/+Z17hEAxt0vt9veFUzMzTcOAH m/1qbxPOHcQ= =auOL -END PGP SIGNATURE-
Preview: request for testers
Attached is a "smart loader" for the previews. I'd really, really like interested users --- that's the mathematicians out there who use /lots/ of equations to test it out. It's meant to enable you to generate and display the previews with no discernible effect on LyX's responsiveness. If you run ./lyx -dbg graphics you'll see what is going on. Having loaded up the document you should be able immediately to start editing it. That's because we already had generation of all these preview images when a document is loaded as a single forked process. You might here ghostscript writing hundreds of image files to disk... One a message like: PreviewLoader::finishedInProgress(0): processing succeeded for sh LYXLIBDIR/scripts/lyxpreview2ppm.sh /tmp/lyx_tmpdir31006aaEiEa/lyx_tmpbuf0/0lyxpreview.tex 111.15 is printed to screen, you'll know that their generation is complete. At present all images are then loaded up into LyX, a process that will effectively hang LyX until all the images are loaded up. Not with this patch. This patch gives you "smart loading" of these images. * The "start loading" request is only sent from the inset's draw method. * Actual loading is then delayed for two seconds, at which point we check whether the inset is actually visible on the screen or not. * If it /is/ visible, then we start loading it. This seems like an elegant solution to me, but I'd like you lot to try it yourselves. If you like the approach, (or rather, if you don't say "I dislike it immensely"), then I'll refactor the code so that the Graphics Inset can do things the same way. No more jumping randomly all over the place as images are loaded. One point to make: the actual implementation still has some bugs, so not everything get's loaded immediately (isInsetVisible sometimes returns false when clearly its true!). What I'm interested in here is whether people like the approach. And if not, whether they can come up with a plausible alternative that does not involve threads. That seems like a solution for the future. One other point (for Jürgen?): I think that a (working) isInsetVisible could be used to advantage by BufferView::updateInset. Seems intelligent to only redraw insets that are visible... Best regards, Angus ps, having applied it, you can always remove this patch with patch -R < smartloading.diff ;-) smartloading.diff.bz2 Description: BZip2 compressed data
