> Does your patch behave differently in that respect?
No. A newer version is trying to reverse empty undo, but it seems
that I have complicated the problem too much.
Bo
>
> > 1. del at [abc|] work as backspace (instead of doing nothing)
>
> No, should melt the inset's content into the surrounding ite,
What exact do you mean? It sounds dangerous and difficult to 'melt the content'.
> > 2. backspace/del at [ [a] [|] [ ] ] moves to the end of first cell
>
> O
> Doing a switch to a different tree than trunk is for me somewhat
> easier than applying a patch. (At least sometimes)
But may not be easy for others, and is not making my life any easier.
I understand that it will make backporting a lot easier if a feature
is committed to the 1.5.x tree as a cor
> You can already do that: branches/abdel/
Why communication becomes so difficult now?
With /branches/bpeng, nobody will check it out and test, except maybe
you, because there are *too many* such branches. With /sandbox or
/next, all potential patches go here and people can play with it and
test
> Attached my next attempt. Now,
pullArg is used at some other places (font something?) Are you sure
your pullArg will work over there?
Bo
> IMO the following procedure would be OK:
Taking the example of the view-source patch,
> If there is agreement about a particular feature,
Yes.
> and also about the
> general implementation (I don't mean the details, but the overall
> approach),
No. The patch is a little big. It has been a li
Dear all,
I check out a fresh lyx trunk, apply my patch, compile. A 'svn diff'
gets me lots of po file changes. Can they be safely committed like
this? Or should I clean them and commit only my patch? (rm -rf po, svn
update; svn commit) Sorry if I am asking something really basic.
Cheers,
Bo
Finally...
> |+void Buffer::getSourceCode(ostream& os, lyx::pit_type par_begin,
> lyx::pit_type par_end)
> |+{
>
> space before (and after) '&' please (same for '*')
> (I see that you are not quite consistent)
Yes. Have double-checked, but ...
> |+ // this is personal. I think source cod
> Or to not specify any specific font, but just ask for the guis
> typewriter font. But hardcoding as it is not is not nice.
I am a total qt newbie and the dialog part is copied from QLog. Can
anyone tell me exactly how to "tell gui use typewriter font', if it is
on the top of your mind?
> | Defi
> Have you fixed J-M concerns about this patch?
Yes. That has been fixed. A painful python regex to boost::regex has been used.
> You are not allowed to use trunk as a place to store your unfinished
> patches/features you know...
This patch is by itself complete. It allows users to open .lyx fil
Dear all,
686 has driven me nuts! I may submit a very simple patch, instead of
trying to fix the whole del/bs scenario.
In the meantime, for your amusement, attached is a small patch that
adds lyx revision number to the about box. This is a quick hack and I
have not though about the whole autogen
On 4/9/06, Andre Poenitz <[EMAIL PROTECTED]> wrote:
> On Sat, Apr 08, 2006 at 03:38:43PM -0500, Bo Peng wrote:
> > In general, I am. I would suggest again that we use a beautifier with
> > one agreed set of parameters.
>
> Do you know one that works with C++?
>
I u
> You are of course aware of svnversion beging a really heavy command
> (as in time consuming)
There is no workaround here but it is also helpful to put revision
number in the about box.
> Also you have to add it to "lyx --version"
Will do.
> | stamp-version: version.C-tmp
> | + sed -i
> > I am a total qt newbie and the dialog part is copied from QLog. Can
> > anyone tell me exactly how to "tell gui use typewriter font', if it is
> > on the top of your mind?
>
> We have lyx_gui::typewriter_font_name() in src/frontends/lyx_gui.h.
Thanks.
> BTW I saw that the handling of OutputPar
> Not all concerns have been fixed. You still use the current directory as
> search path. This should be only the document directory.
Where? You mean the default parameter of load(path=".")? If I use
something like load(path=""), The code will try to locate
"/file.layout", and of course we do not
> There is no need for the cast in C (as the conversion is automatic),
I have not checked C99 standard on this issue, but C has improved on
such things and may no longer allow this dangerous implicit
conversion.
> and there is no need for malloc in C++ (as there's 'new' etc).
Right. Poeple still
> Yes. I originally intended to wait with introducing this till we
> changed to xml, but I think it might be useful even now.
I am not qualified for an opinion in this area, just to say that all
these pit, id thing have confused me. I hope this introduction of uuid
will make things easier, not har
> PS: I already use the session features and really like it that LyX starts
> with the last file open. I would not have thought that before actually
> using it!
I am happy to know that. Keep your view-source dialog open, you may
like that as well. Note that change-track -> view output with change
> Well, I used it daily so I care. If you think this is not worth it, I
> will just stop working on that.
Please do not. I will be interested in using it myself. After qt4 is
at a comparable state with qt2. I am also interested in adding syntex
coloring of my view source dialog for latex outputs.
> | My version of svnversion requires a path anyway. The attached patch works
> | for me perfectly with srcdir != builddir,and it forces recreation of the
> | version info for every build.
This is good, and is the expected behavior. (update, modify, compile,
update, compile. get new revision numbe
> ii) Is it possible to let LyX remember its wondow size and position from the
> last session?
This, along with other session-related features, has been implemented
and is under testing. If you repeat this request 100 times :-), you
may see this feature in the coming lyx 1.4.1. Otherwise, I am su
Dear all,
Attached is the second patch of the view-source feature, with changes
according to Lars' and Georg's suggestions.
The most significant change is the dialog now updates itself with any
update event. This means, at any mouse click, selection, key entry,
the opened view-source dialog will
> So feel free to revoke my svn access. Anyway, no need to discuss that
> further more I guess main devels won't change their mind.
Applying patches from list, or switching to another branch, or having
a next branch, have all proven to be a big burden to the lyx veterans.
I am taking a more aggres
> Angus has suggested that someone creates his own branch and work from
> there.
My understanding is that branches are for big changes, not for single
small features like mine. Even if you are willing to switch to it,
others may not. Also, keeping a branch up to date is a big burden, and
merging t
> Seems that your "timely" manner is completely different from ours.
Give me a time frame you think it is reasonable for me to wait.
> | I am taking a more aggressive two/three-step approach now. Namely I
> | will apply the patch (post to the list first), and try to make it
> | right with the sec
> And for the harsher demands on the users system. I guess that can be
> rectified by setting up a callback mechanism from the core ->
> frontend.
That is what my first patch does: updating the dialog only when the
core asks, but some sort of callback mechanism should be better.
Regarding the QFo
> The cursor positioning, although useful, is also a bit lacking. If the
> cursor is left inside an inset when quitting lyx, it will be placed in
> a "random" spot when starting up again.
This is what internal bookmarks' behavior, right? I do not see where
bookmarks save the whole cursor stack.
W
Dear all,
When I have the view-source dialog on, I sometimes (very difficult to
reproduce so I have not caught it in gdb) see lyx crash with
Assertion triggered in Point CoordCache::get(const LyXText*,
lyx::pit_type) by failing check "posit != it->second.end()" in file
coordcache.C:64
Abort
The
> Bo> What I can offer is to store the cursor position of the top-level
> Bo> paragraph. When restoring an article, the cursor will be before an
> Bo> inset, instead of in the exact location in the inset when the
> Bo> document was closed. I think this is enough for this feature.
>
> What is so dif
> Your first patch really had the core update the dialog didn't it?
First patch: The core prepare the text, send to the dialog for display.
Second patch: The dialog update itself, for whatever update event.
Optimal way: the core feels the need to update the dialog, send a
signal, and the dialog
Dear all,
Playing with the view-source dialog (it is great fun), I can not help
thinking whether or not we can achieve something similar to
dreamweaver's dual view feature. That is to say, what is typed in the
lyx window shows automatically (and dynamically) in the source window
(has been done), a
OK, so a few people are interested.
If we limit the use of such feature, at first, only to math. This
feature is doable. Remember that if we have a $x^2$ as text, select
and press C-M, we can get x^2 in math mode. This can be used in this
case.
1. in the view-source (edit-source later?) dialog, i
Dear list,
With beamer class, under 1.4.x and 1.5svn, the enumerate environment
does not show numbers before the items. Itemize environment works
fine.
What should I add to beamer.layout?
Cheers,
Bo
> > What should I add to beamer.layout?
>
> Nothing. Try the attached patch.
>
The patch works for me.
Thanks.
Bo
> That would look like the attached. Does it now do the right thing?
yes.
Bo
Dear list,
I think this is a known problem, since I have heard (misheard) a few
words about it.
In beamer, when I delete one or multiple frames, the skeleton of the
deleted part will remain. That is to say, the frame, columns, item ...
environments will remain, although all texts are removed. I
Dear list,
1. Open a document, edit the first paragraph
2. Scroll down a few pages, using mouse or page down key
3. File->save
4. Lyx will scroll back to the beginning of the document.
Since I use Alt-F,S very often (a psychological result of previous
disastrous data loss, not caused by lyx) this
> Yes. Even better: find out why this happens (no I don't have an idea) ;-)
> BTW, I did not notice that so far.
Really? This happens on Linux/RHEL4/qt2/lyx1.5.xsvn. I will see what I can find.
There has been some other minor cursor movement problems. For example,
sometimes, I just can not click
QViewSource.C (revision 0)
+++ src/frontends/qt4/QViewSource.C (revision 0)
@@ -0,0 +1,110 @@
+/**
+ * \file QViewSource.C
+ * This file is part of LyX, the document processor.
+ * Licence details can be found in the file COPYING.
+ *
+ * \author John Levon
+ * \author Bo Peng
+ *
+ * Full a
> If your Qt4 version works i.e. compile and doesn't make lyx crash,
> please commit it yourself and we work from there.
Done.
The ugliness of qt4 comes from the fact (I guess) that my kde3 uses
qt3 and qt4 can not get a working style file from the system. I guess
I can get such things from onlin
> looks ok here
>
> general comment: i think "view latex code" or something like that would
> be more appropriate
>
This dialog can view other source type like linuxdoc and docbook, so
it is better termed as 'view-source'. There is currently one type of
syntax highlighting though.
Bo
On 4/15/06, Georg Baum <[EMAIL PROTECTED]> wrote:
> Am Samstag, 15. April 2006 19:23 schrieb Bo Peng:
> > This dialog can view other source type like linuxdoc and docbook, so
> > it is better termed as 'view-source'. There is currently one type of
> > syntax h
> First, the lyx installer looked up the websites of all these
> other necessary things (gostscript, pythong,...)
> Unfortunately, none were actually installed, I just got
> the correct download page in explorer. But that's
> ok, I new what to do although it'd be ice if it was automatic.
Uwe, peo
With this checkbox, user can switch between viewing the whole source
code, and part of it (current behavior). This is useful to view the
front matters.
Since automatic update of the whole source code may become too much a
burden, I propose an additional 'auto update' checkbox which will be
turned
Hi, Uwe,
Is scons/msvc working for you now? If not, could you please give me
the exact command line and error messages?
Bo
Quote from Status.15x:
TOOLBARS & MINIBUFFER
* When toolbars are hidden, they pop up again after clicking on the document
area (Joost 4/11/06).
I do not observe this. Is this windows only?
* M-x is broken (does not bring up the minibuffer).
So what exactly do we want M-x to do?
* Command co
> * M-x is broken (does not bring up the minibuffer).
>
> So what exactly do we want M-x to do?
It should make the minibuffer visible and focus on it (but not turn it
on permanently).
You mean hide the toolbar as soon as the user hit enter (and execute
the command)? This does not sound right.
Bo> I tend to suggest that we remove this feature.
I think we should not. Some people actually rely on this feature, and
we would not gain much by removing it.
OK. I admit that I am not a power user, at least regarding this
minibuffer feature. The problem now is that M-x is supposed to work
eve
Having the minibuffer as tollbar is ugly and eats screen real estate for
no good reason. Should be merged with the status line.
I agree, but having a editable status line is kind of peculiar.
The minibuffer is not accessible by M-x when no buffer is loaded.
I have proposed to remove this sho
Hmm it's close to pointless without being able to switch to it easily. I
don't understand why it's difficult?
Becasue this feature is rarely used by normal user so the minibuffer
toolbar should not be shown by default. Then, M-x will fail if it can
not find minibuffer. I propose that power users
> What prevents it from doing what it already did in 1.4?
There has been some toolbar reorganization that is not yet finished. I have
no idea if this is a big problem or not.
The problem is that M-x does not always work.
Bo
On 11/18/06, John Levon <[EMAIL PROTECTED]> wrote:
See below. This makes the shown minibuffer as part of the session, but I
don't think that's problematic. There are still other problems:
Cannot you detect minubuffer toolbar state and call toggleToolbarState
if needed? You can avoid setToolbar
I do not really like the idea of showing minibuffer with M-x but it
may be better than getting rid of M-x altogether, at least for power
users.
I have two patches: The first is needed anyway, and fixes the problem
so I applied it. The second patch removes M-x. It has no chance of
getting approve
> Kornel Benko wrote:
> > utf8ToUcs4 is defined in lyx_main.o, but this modul is not used to bind
> > lyxclient.
Dear all,
Can I propose that we pay more attention to our patches and make sure
that lyx (including tex2lyx and lyxclient) can compile before
submitting them? Uncompilable trunk is f
http://bugzilla.lyx.org/show_bug.cgi?id=686
--- Additional Comments From [EMAIL PROTECTED] 2006-11-18
22:50 ---
I would also like to see this feature, i.e, delete empty math box with
delete/backspace key. It is not working now. If I use the delete or backspace
keys within an empty math b
(Yes, I strongly think we should keep M-x as a shortcut. Removing it
just because the fix for making M-x always work... well, that sucks.).
M-x now works. The side effect is that it turns on minibuffer toolbar,
which has to be turned off manually.
Bo
No, I still get the same error message.
To reproduce follow the steps given in trunk in the file
"INSTALL.Win32". The build script I used as adviced there is
~ development\Win32\packaging\build_msvc.bat
I can not repeat your problem. I get python 2.5, scons 0.96.93 (I used
python 2.4 and scons 0
Hi, Peter,
Does cmake define HAVE_CHMOD? Scons detects chmod under msvc and
defines HAVE_CHMOD. However, in src/support/copy.C, chmod is used with
mode_t which does not exist under windows.
HAVE_MODE_T may be needed.
Bo
> Does cmake define HAVE_CHMOD?
No, after discussing it with Georg I've disabled chmod on Windows,
config.h.cmake:
#ifdef _MSC_VER
I see. I am uncomfortable with #_MSC_VER since this assumes msvc
compiler so this may fail under mingw.
I am adding HAVE_MODE_T. Please, Georg and Peter, adjust a
> mode_t is only used when HAVE_CHMOD is enabled
> so why do you need a HAVE_MODE_T?
Because msvc has chmod, but not mode_t; and I am reluctant to use
compiler/os specific flags.
And mingw has a mode_t.
msvc is only *one* compiler under windows. mingw happens to have
mode_t, but what about c
thanks for the hint but I had the SDK already installed. I followed
exactly the advices from the link above but still got the same error
message. After fiddling around some hours :-(, and reading newsgroups I
created two new Win environment variables and now it works.
I do not quite understand w
I just spent an hour on a xr related bug/problem. Since this may be
encountered by others, please allow me to explain here for future
reference.
In docA.tex, a label is usually presented in the aux file as
\newlabel{ref}{{1}{1}}
and the xr package in docB can process it correctly.
However, if \u
Next time having trouble have also a look at /development/cmake/README.cmake:
- add include and library paths of the SDK to the IDE search paths,
menu: Tools->Options->VC++ directories->Library files + Include files
Is not this only for the IDE? I mean, after the change, will the vc
command p
Sorry Ugras, but I don't have much time today to test your patch. I
suggest that you send again the full patch. Then, maybe someone else can
test and commit for you (Edwin, Bo, Georg?). If not, I'll try to do that
at the end of the week.
I heard my name being called. OK, I will try this patch du
> and there is a \newpage ...
I know. I always thought that it was equivalent to \clearpage. If that is
not true, what is the difference?
"The \clearpage command ends the current page and causes all figures
and tables that have so far appeared in the input to be printed."
\newpage only starts a
Thanks, Bo. Juergen, can you also test it, to see if it is enough for
your taste? At the end, you were the one who asked for the feature.
I am not sure if this exists before the patch, but when I click on any
item, it moves to the end of the dialog. This is very distracting
since I have to re-re
If you back this feature as well that makes three developers and the code
can be committed.
I like this feature and read the patch briefly. I am not sure what you
mean by 'back', but I guess I can fix problems of this feature when
bugs are found (not tex2lyx or lyx2lyx though).
Bo
Good, then apply it ;-)
The patch looks simple but I do not know that part of the code. Please
evaluate the implementation, modify and apply it. It would be best if
the 'click and move' problem Juergen and I noticed can be corrected.
Cheers,
Bo
Maybe at some point, everybody should align on one set of building tools
(autotools, scons or cmake) and that we have a clear, regular and
documented for dummies path to build LyX across platforms and Linux
distributions.
The documentation of scons is currently the best :-). INSTALL.Win32 by
Joo
} else if (token == "\\newpage") {
par.insertInset(par.size(), new InsetPagebreak, font, change);
should be InsetNewpage
I do not quite get it. You mean the original code for newpage is broken?
Bo
On 11/23/06, Edwin Leuven <[EMAIL PROTECTED]> wrote:
2>c:\program files\gnuwin32\include\zconf.h(289) : fatal error C1083:
Cannot open include file: 'unistd.h': No such file or directory
Is this mingw/windows/scons, or anything else?
Bo
Attached is what I believe a cleaner implementation of
clearpage/cleardoublepage. Please, tex2lyx and lyx2lyx experts, fix
tex2lyx and lyx2lyx so that I can apply it.
Thanks.
Bo
Index: src/LyXAction.C
===
--- src/LyXAction.C (revisio
Sigh. Once again: tex2lyx _must not_ be changed unless somebody implements
the changes from format 245 to 254 first. Otherwise tex2lyx will generate
invalid files.
I was asking if someone can complete this patch with needed tex2lyx
changes so that this patch will *not* produce invalid files. Wha
If you put this in I'll test the lyx2lyx part and fix it if needed.
Done. latexLabel changed to getCmdName. Thanks.
Bo
I get this message every time I start LyX:
LyX: Warning: Ignore last opened file: /home/software/newfile1.lyx
LyX: Warning: Ignore pos of last file: /home/software/newfile1.lyx
It is strange to me that newfile1.lyx is there everytime since ignored
files will not be saved. If you start lyx,
BTW: I find it quite confusing that clearpage is derived from pagebreak.
There should be some "empty" base class from which all three command
classes are derived.
clear page etc extend page break so the logic is not so strange here.
Anyway, using pagebreak and its derived class as an example, i
last and very IMPORTANT: I have sent a better patch in a separate mail
including a new feature "\linebreak", which was also asked in the same
mail that clearpage asked. I will be glad if you people review that
patch, be volunteer to support this feature (so, Jose can put it for
alpha), and commit,
No, as Martin said, so much has changed in cache handling in 1.5 that it
would be a lot of work to try to correct this in 1.4. This would
destabilize 1.4 way too much I'm afraid.
I have a feeling that if we continue to improve 1.5.x like this, we do
not need any 1.4.5...
It would even make sens
Hi, Peter,
icons size is now stored in session, but is there a way to set icon size?
Bo
I do not understand it either, other than the fact that José wants to
rewrite everything in python.
I also want to rewrite everything in python. :-)
Seriously, tex2lyx is rarely used (so speed is not an issue) and
involves a lot of string handling. Python seems to be a much better
language for
Hi, Abdel,
The buttons in the lyx documents are eye-catching (distracting) and ugly
(personal view) so I prefer to make them less prominent using
\set_color "insetbg" "#cc2263"
\set_color "top" "#5500ff"
\set_color "bottom" "#5500ff"
\set_color "left" "#5500ff"
\set_color "right" "#5500ff"
\set_
This is because normal mouse notion (i.e. without any button pressed) is
not processed in GuiWorkarea.
Thanks. I need to add setMouseTracking(true) in
GuiWorkArea::GuiWorkArea. Before I investigate further, do you have
any idea about the performance impact of this? Basically, I need to
figure ou
You'll need to be a little cleverer than that to ensure that v remains in the
range 0 <= v <= 1, but that's all. There may well be code to do all that
already, but my memory fades.
The look of button is fine (to me) after tweaking. A redesign of how
they are drawn is of course possible.
My poin
Abdelrazak Younes wrote:
> Good news for Windows and Mac users, this bug is now fixed:
Another good news is that I am now at 13 seconds for the UserGuide
scrolling test :-)
Congratulations and many thanks. Now you need to improve the
performance of multiple windows. I have been working on two d
What about using short ints instead of ints?
I do not know what exactly is going on here, but if there are a lot of
duplicates, std::map should help.
Bo
Dear all,
Please find attached a patch that
1. Enable mouse movement tracking in qt4/GuiWorkArea.C
2. insetbase.h/C: give inset has_focus_ property so that insets can
choose to draw differently w.r.t this flag.
3. in src/BufferView.C, set has_focus_ for insets and refresh the
paragraph when mo
Dear all,
The attached patch will be applied tomorrow morning if there is no
objection. The main design ideas are that
1. enable mouse tracking
2. get inset under mouse (done by editXY) and pass notifyMouseEnter
and notifyMouseLeave to it.
3. if an inset (currently insetcommand and insetcollapsab
IIUC, this is not a signal but a call to process some mouse event so
'dispatchMouseEntered()' or 'processMouseEntered()' or even
mouseEnteredEvent() all seem more meaningful to me.
I will combine the two functions as suggested.
> +bool InsetCollapsable::notifyMouseEnter()
> +bool InsetCollapsa
Well, in practice, this is the same as most insets are 'clickable'.
'clickable' is a wrong word. The repaint should only happen for
collapsable insets (like footnote) and commands (like toc). They are
relatively rare compared to text and mathed. Also, moving inside an
inset should not trigger an
I don't think this will happen in 1.5 so, as I am sure there would be a
problem with Mac, I'd say that you should either make this new feature
configurable or disabled on Mac.
I still think the number of repaints would be small. Let us see what
Bennet's test result is.
.
Bo
An updated patch has just been submitted. Enjoy!
Bennett: if this patch poses a severe performance problem for mac, I
will disable mouse-tracking for mac.
How about setMouseHover? setMouseIn(false) sounds funny.
setMouseHover is used.
> 4) The toggling does not work for opened insets, only
Dear all,
Mouse-hover over a button triggers a full repaint, instead of the
single paragraph repaint that the code demands. I had a look at the
code:
1. mouse_motion returns (true, true) when a single paragraph need to
be updated. (BufferView.h/C)
/// \sa WorkArea
/// \retval (f
> Note that partial repaint should be implemented regardless of my
> button patch.
Partial repaint works fine without your patch ;-)
So you mean I can not request partial redraw for the paragraph with
hover-style buttons? Then, the comment of workAreaDispatch needs to be
updated.
/// \sa
Bo, I don't like this new look. I like to be able to easily locate
all the insets and now this is not possible anymore.
The only thing you need to do is change \buttonbg to some other darker
color, from the preference dialog. This color is currently set to text
background color. It is also possi
Bo, I don't like this new look.
I know it is hard to reach a consensus, but these insets (reference,
footnotes etc) contain auxiliary information, and should not catch
more attention than the main text. Also, as I have explained, buttons
in a lyx document are not like buttons in a html page. In
> But anyway, the colors are configurable...
Ok, I'll try that.
I guess thicker frame plus a more prominent frame color (purple?) will
fit your need? I am open to such changes to make these buttons appeal
to 80% of the users (if we are at 50% now, and 100% is not possible).
Bo
1) Alt + right arrow: if the keyboard cursor is immediately right to a
collapsed inset, this inset will be automatically opened and the cursor
positioned at the beginning of the inset.
2) Alt + left arrow: if the keyboard cursor is immediately left to a
collapsed inset, this inset will be automat
> Actually, I achieve the desired goal by making a little darker the
> button background ;-)
If the button looks fine there, could you send me the color you use?
BTW: Would it be too demanding if I ask for rounded boxes?
It should not be too difficult. You can try to do that in painter.C,
an
http://www.lyx.org>
We are actually quite close to these. The buttons, with a background
identical to text background, look like regular text. And we can tweak
individual insets to what it displays.
If you are unhappy about the label of a certain inset, we can discuss
and change it.
Cheer
On 12/4/06, Georg Baum <[EMAIL PROTECTED]> wrote:
Please fix this warning:
../../../src/mathed/../insets/insetbase.h:185: warning: unused
parameter 'mouse_hover'
Fixed. Thanks.
Bo
901 - 1000 of 4033 matches
Mail list logo
