Discussion:
Str-metrics branch: can I land now?
(too old to reply)
Jean-Marc Lasgouttes
2014-07-25 08:49:35 UTC
Permalink
Hello,

The str-metrics branch has come to a point where I thing it should be in
master. I have two more commits at home (forgot to push) that are going
to solve the remaining cursor positioning problem that was haunting me.

I do not feel that it gets much testing now, and I am going to leave in
vacation on wednesday. Therefore in some sense it is now or never :)

Seriously, does anyone have some reservations against merging this work
to branch now?

JMarc
Jean-Marc Lasgouttes
2014-07-25 08:50:25 UTC
Permalink
Post by Jean-Marc Lasgouttes
Seriously, does anyone have some reservations against merging this work
to branch now?
Of course, I will rewrite history a last time before doing that.

JMarc
Kornel Benko
2014-07-25 09:27:45 UTC
Permalink
Post by Jean-Marc Lasgouttes
Hello,
The str-metrics branch has come to a point where I thing it should be in
master. I have two more commits at home (forgot to push) that are going
to solve the remaining cursor positioning problem that was haunting me.
I do not feel that it gets much testing now, and I am going to leave in
vacation on wednesday. Therefore in some sense it is now or never :)
I only compile it from time to time. The version from yesterday compiles fine.
Sorry for not testing.
Post by Jean-Marc Lasgouttes
Seriously, does anyone have some reservations against merging this work
to branch now?
I am for merging it. The longer it lasts, the more work is to be done IMHO.
Post by Jean-Marc Lasgouttes
JMarc
Kornel
Jean-Marc Lasgouttes
2014-07-25 09:47:52 UTC
Permalink
Post by Kornel Benko
Post by Jean-Marc Lasgouttes
Seriously, does anyone have some reservations against merging this work
to branch now?
I am for merging it. The longer it lasts, the more work is to be done IMHO.
I think the remaining problems may be for bidi text, but I am not sure
how many testers/developers we have for bidi.

One point which is still open is the handling of delimiters. Currently
we have weird code that turns ( to ) in RTL code on display, but
differently for hebrew and arabic. The branch changes the way these
delimiters are displayed (this is part of the unicode bidi algorithm),
but this may not be right. However, this stuff needs to be cleaned up,
which probably requires a file format change, at least for hebrew.

I am not going to look at that. I'll wait until some rtl user complain
about the delimiters display and see with them what solution has to be used.

JMarc
Jürgen Spitzmüller
2014-07-25 11:24:17 UTC
Permalink
Post by Jean-Marc Lasgouttes
Hello,
The str-metrics branch has come to a point where I thing it should be in
master. I have two more commits at home (forgot to push) that are going
to solve the remaining cursor positioning problem that was haunting me.
I do not feel that it gets much testing now, and I am going to leave in
vacation on wednesday. Therefore in some sense it is now or never :)
Seriously, does anyone have some reservations against merging this work
to branch now?
Merge it.

Jürgen
Post by Jean-Marc Lasgouttes
JMarc
Richard Heck
2014-07-25 14:06:46 UTC
Permalink
Post by Jürgen Spitzmüller
Post by Jean-Marc Lasgouttes
Hello,
The str-metrics branch has come to a point where I thing it should be in
master. I have two more commits at home (forgot to push) that are going
to solve the remaining cursor positioning problem that was haunting me.
I do not feel that it gets much testing now, and I am going to leave in
vacation on wednesday. Therefore in some sense it is now or never :)
Seriously, does anyone have some reservations against merging this work
to branch now?
Merge it.
I agree.

rh
Kornel Benko
2014-08-02 12:38:20 UTC
Permalink
Post by Richard Heck
Post by Jürgen Spitzmüller
Post by Jean-Marc Lasgouttes
Hello,
The str-metrics branch has come to a point where I thing it should be in
master. I have two more commits at home (forgot to push) that are going
to solve the remaining cursor positioning problem that was haunting me.
I do not feel that it gets much testing now, and I am going to leave in
vacation on wednesday. Therefore in some sense it is now or never :)
Seriously, does anyone have some reservations against merging this work
to branch now?
Merge it.
I agree.
rh
Now that the merging is done, I see a different behaviour in entering number keys (the ones
at the main keyboard, not the numeric ones)

Entering the string "1234567890" gives me "+ğščťşÜáíé".
(But it is OK on the normal console)

The GUI is slovak, and this output would be OK, if the were a slovak keyboard, which is not.

It does not depend, if I change the gui document language.

While having set lyx -dbg key,..:
Entering e.g. '9' leads to

...
frontends/qt4/GuiWorkArea.cpp (1075): count: 1 text: 9 isAutoRepeat: 0 key: 57 keyState:
frontends/qt4/GuiKeySymbol.cpp (623): Getting key 57, with text '9'
frontends/qt4/GuiKeySymbol.cpp (629): Setting key to 57, 9
frontends/qt4/GuiKeySymbol.cpp (644): isOK is 1
frontends/qt4/GuiApplication.cpp (2019): KeySym is 9
frontends/qt4/GuiKeySymbol.cpp (644): isOK is 1
frontends/qt4/GuiKeySymbol.cpp (652): isModifier is 0
frontends/qt4/GuiApplication.cpp (2042): action first set to [self-insert]
frontends/qt4/GuiApplication.cpp (2050): action now set to [self-insert]
frontends/qt4/GuiApplication.cpp (2061): Key [action=self-insert][9]
frontends/qt4/GuiApplication.cpp (1494): cmd: action: 84 [self-insert] arg: '9' x: 0 y: 0
BufferView.cpp (1219): BufferView::dispatch: cmd: action: 84 [self-insert] arg: '9' x: 0 y: 0
insets/InsetText.cpp (267): InsetText::doDispatch(): cmd: action: 84 [self-insert] arg: '9' x: 0 y: 0
Text3.cpp (454): Text::dispatch: cmd: action: 84 [self-insert] arg: '9' x: 0 y: 0
BufferView.cpp (2590):
y1: 0 y2: 480 pit: 0 singlepar: 1
...
Paragraph.cpp (4030): misspelled word: "ğščťşÜáí" [1..9]
...
frontends/qt4/LayoutBox.cpp (558): Already had Plain Layout selected.
frontends/qt4/GuiApplication.cpp (1283): dispatch msg is
frontends/qt4/GuiApplication.cpp (2102): SelfInsert arg[`9']


Kornel
Jean-Marc Lasgouttes
2014-09-10 08:22:32 UTC
Permalink
Post by Kornel Benko
Now that the merging is done, I see a different behaviour in entering number keys (the ones
at the main keyboard, not the numeric ones)
Entering the string "1234567890" gives me "+ľščťžýáíé".
(But it is OK on the normal console)
The GUI is slovak, and this output would be OK, if the were a slovak keyboard, which is not.
It does not depend, if I change the gui document language.
Hi Kornel,

Has there been any progress on this? The str-metrics branch does not
touch keyboard input at all, only on-screen rendering. I am very
surprised that it could cause what you describe here.

JMarc
Kornel Benko
2014-09-10 09:14:35 UTC
Permalink
Post by Jean-Marc Lasgouttes
Post by Kornel Benko
Now that the merging is done, I see a different behaviour in entering number keys (the ones
at the main keyboard, not the numeric ones)
Entering the string "1234567890" gives me "+ğščťşÜáíé".
(But it is OK on the normal console)
The GUI is slovak, and this output would be OK, if the were a slovak keyboard, which is not.
It does not depend, if I change the gui document language.
Hi Kornel,
Has there been any progress on this? The str-metrics branch does not
touch keyboard input at all, only on-screen rendering. I am very
surprised that it could cause what you describe here.
JMarc
I don't know, what caused this. Cannot reproduce anymore.
(I remember seeing it for a week or so. Since it was on a virtual
testing machine, It may as well have been some unusual configuration)

It is working well since then, so I already forgot the issue.

Kornel
Kornel Benko
2014-10-13 08:16:46 UTC
Permalink
Post by Kornel Benko
Post by Jean-Marc Lasgouttes
Post by Kornel Benko
Now that the merging is done, I see a different behaviour in entering number keys (the ones
at the main keyboard, not the numeric ones)
Entering the string "1234567890" gives me "+ğščťşÜáíé".
(But it is OK on the normal console)
The GUI is slovak, and this output would be OK, if the were a slovak keyboard, which is not.
It does not depend, if I change the gui document language.
Hi Kornel,
Has there been any progress on this? The str-metrics branch does not
touch keyboard input at all, only on-screen rendering. I am very
surprised that it could cause what you describe here.
JMarc
I don't know, what caused this. Cannot reproduce anymore.
(I remember seeing it for a week or so. Since it was on a virtual
testing machine, It may as well have been some unusual configuration)
It is working well since then, so I already forgot the issue.
The behaviour is back. It has something to do with my preferences.

If I use a fresh userdir, then there is no such effect. Copying the preferences over my own, the effect
vanishes.

It is, oh wonder,
\kbmap true
\kbmap_primary "slovak"

So, problem solved.

Kornel
Jean-Marc Lasgouttes
2014-10-13 09:21:36 UTC
Permalink
Post by Kornel Benko
If I use a fresh userdir, then there is no such effect. Copying the preferences over my own, the effect
vanishes.
It is, oh wonder,
\kbmap true
\kbmap_primary "slovak"
So, problem solved.
Very good :)

JMarc

Loading...