Loading ...
Sorry, an error occurred while loading the content.

Re: Weird behavior after ^ or ´

Expand Messages
  • Axel
    Hi Ben, reproducible with both calls here. -- You received this message from the vim_dev maillist. Do not top-post! Type your reply below the text you are
    Message 1 of 24 , Oct 29, 2012
    • 0 Attachment
      Hi Ben,

      reproducible with both calls here.

      --
      You received this message from the "vim_dev" maillist.
      Do not top-post! Type your reply below the text you are replying to.
      For more information, visit http://www.vim.org/maillist.php
    • Christian Brabandt
      ... And how do you start Vim? Did you test with vim -u NONE -U NONE -N Does cl work like s (it should, since s is an alias to cl)? regards, Christian -- You
      Message 2 of 24 , Oct 29, 2012
      • 0 Attachment
        On Mon, October 29, 2012 16:17, Axel wrote:
        > "s" makes the cursor move to the beginning of the line(!) changing the
        > character found there (hence the "$" sign). Also, the character "^" is in
        > fact not displayed.

        And how do you start Vim? Did you test with vim -u NONE -U NONE -N
        Does cl work like s (it should, since s is an alias to cl)?

        regards,
        Christian

        --
        You received this message from the "vim_dev" maillist.
        Do not top-post! Type your reply below the text you are replying to.
        For more information, visit http://www.vim.org/maillist.php
      • Axel
        Addendum: Sorry all, as Ben has already noticed, this is a double post. The original one is in the place he indicated... I ll switch over to the other one. --
        Message 3 of 24 , Oct 29, 2012
        • 0 Attachment
          Addendum: Sorry all, as Ben has already noticed, this is a double post. The original one is in the place he indicated...

          I'll switch over to the other one.

          --
          You received this message from the "vim_dev" maillist.
          Do not top-post! Type your reply below the text you are replying to.
          For more information, visit http://www.vim.org/maillist.php
        • Jürgen Krämer
          Hi, ... I can confirm this behavior on Vim 7.3.1-712 on Windows 7, compiled with MS-C 16.0.40219.1 (i.e., Visual Studio 2008), although it only seems to happen
          Message 4 of 24 , Oct 29, 2012
          • 0 Attachment
            Hi,

            Ben Fritz schrieb:
            > On Monday, October 29, 2012 9:34:36 AM UTC-5, Christian Brabandt wrote:
            >> On Mon, October 29, 2012 11:21, Axel wrote:
            >>
            >>> Since some patches (right now at 7.3.712) I experience the following
            >>> behavior ([] denotes the cursor position, the character after the arrow
            >>> the input):
            >>>
            >>> a[a]aa -> a
            >>> aa[]aa -> <spc>
            >>> aa []aa -> ^
            >>> aa []aa -> <spc> " Should result in "aa ^[]aa"
            >>> aa []aa -> <esc>
            >>> aa[ ]aa -> s " gA shows 0x20
            >>> []$a aa " Should result in "aa[]$aa"
            >>>
            >>> This seems to be a bug. Can anyone verify this (Windows 7 64 bit;
            >>> MinGW-compiled (64-bit))?
            >>>
            >>> Addendum: This also seems to happen with apostrophes.
            >>
            >> I am not sure, I understand your description, e.g. What are you
            >> typing after the s key (possibly the dollar)? Is Vim really not
            >> showing the caret (^) after the space? Why do you have a $ sign there?
            >>
            >> And I can't reproduce this issue on Windows (but 32bit, Cream installer)
            >
            > That is 3 people now who cannot reproduce. I'm not sure why the OP started a new thread. Here's the original, to my knowledge:
            > https://groups.google.com/d/topic/vim_use/3MUcDw_N94Q/discussion
            > "Axel", can you reproduce this without any of your plugins or .vimrc? Try after starting Vim like gvim -N -u NONE -i NONE -U NONE. If the problem goes away also try with gvim --noplugin -i NONE to figure out whether it's one of your plugins.

            I can confirm this behavior on Vim 7.3.1-712 on Windows 7, compiled with
            MS-C 16.0.40219.1 (i.e., Visual Studio 2008), although it only seems to
            happen of every other input of "^".

            The caret is a dead-letter key and is ignored although a space is pressed
            afterwards. It seems to be kept in the input buffer, though, and is
            finally used when "s" is pressed, leading to a movement to the start of
            line and starting insert mode at the wrong position.

            The dollar sign you see in Alex' example is the one displayed at the end
            of the changed text if "$" is included in 'cpo'.

            Regards,
            Jürgen

            --
            Sometimes I think the surest sign that intelligent life exists elsewhere
            in the universe is that none of it has tried to contact us. (Calvin)

            --
            You received this message from the "vim_dev" maillist.
            Do not top-post! Type your reply below the text you are replying to.
            For more information, visit http://www.vim.org/maillist.php
          • Jürgen Krämer
            Hi, ... this behavior does not depend on using cl or s. The bug lies in ignoring the caret in insert mode and using it later before executing the next command.
            Message 5 of 24 , Oct 29, 2012
            • 0 Attachment
              Hi,

              Christian Brabandt wrote:
              > On Mon, October 29, 2012 16:17, Axel wrote:
              >> "s" makes the cursor move to the beginning of the line(!) changing the
              >> character found there (hence the "$" sign). Also, the character "^" is in
              >> fact not displayed.
              >
              > And how do you start Vim? Did you test with vim -u NONE -U NONE -N
              > Does cl work like s (it should, since s is an alias to cl)?

              this behavior does not depend on using cl or s. The bug lies in ignoring
              the caret in insert mode and using it later before executing the next
              command. It seems that only every other (dead-letter) caret is inserted
              even if it is followed by a space.

              Regards,
              Jürgen

              --
              Sometimes I think the surest sign that intelligent life exists elsewhere
              in the universe is that none of it has tried to contact us. (Calvin)

              --
              You received this message from the "vim_dev" maillist.
              Do not top-post! Type your reply below the text you are replying to.
              For more information, visit http://www.vim.org/maillist.php
            • Jürgen Krämer
              Hi, ... additional note: I could only reproduce this in GVim. Regards, Jürgen -- Sometimes I think the surest sign that intelligent life exists elsewhere in
              Message 6 of 24 , Oct 29, 2012
              • 0 Attachment
                Hi,

                Jürgen Krämer wrote:
                >
                > I can confirm this behavior on Vim 7.3.1-712 on Windows 7, compiled with
                > MS-C 16.0.40219.1 (i.e., Visual Studio 2008), although it only seems to
                > happen of every other input of "^".
                >
                > The caret is a dead-letter key and is ignored although a space is pressed
                > afterwards. It seems to be kept in the input buffer, though, and is
                > finally used when "s" is pressed, leading to a movement to the start of
                > line and starting insert mode at the wrong position.
                >
                > The dollar sign you see in Alex' example is the one displayed at the end
                > of the changed text if "$" is included in 'cpo'.

                additional note: I could only reproduce this in GVim.

                Regards,
                Jürgen

                --
                Sometimes I think the surest sign that intelligent life exists elsewhere
                in the universe is that none of it has tried to contact us. (Calvin)

                --
                You received this message from the "vim_dev" maillist.
                Do not top-post! Type your reply below the text you are replying to.
                For more information, visit http://www.vim.org/maillist.php
              • Ben Schmidt
                ... Caret is not a dead letter for me. So is this bug locale-dependent? Ben. -- You received this message from the vim_dev maillist. Do not top-post! Type
                Message 7 of 24 , Oct 29, 2012
                • 0 Attachment
                  On 30/10/12 2:40 AM, Jürgen Krämer wrote:
                  > Hi,
                  >
                  > Christian Brabandt wrote:
                  >> On Mon, October 29, 2012 16:17, Axel wrote:
                  >>> "s" makes the cursor move to the beginning of the line(!) changing the
                  >>> character found there (hence the "$" sign). Also, the character "^" is in
                  >>> fact not displayed.
                  >>
                  >> And how do you start Vim? Did you test with vim -u NONE -U NONE -N
                  >> Does cl work like s (it should, since s is an alias to cl)?
                  >
                  > this behavior does not depend on using cl or s. The bug lies in ignoring
                  > the caret in insert mode and using it later before executing the next
                  > command. It seems that only every other (dead-letter) caret is inserted
                  > even if it is followed by a space.

                  Caret is not a dead letter for me. So is this bug locale-dependent?

                  Ben.



                  --
                  You received this message from the "vim_dev" maillist.
                  Do not top-post! Type your reply below the text you are replying to.
                  For more information, visit http://www.vim.org/maillist.php
                • Christian Brabandt
                  Hi Ben! ... It is for a German layout usually. But I can t reproduce it. And possibly also compiler or architecture (32/64bit) dependent. Mit freundlichen
                  Message 8 of 24 , Oct 29, 2012
                  • 0 Attachment
                    Hi Ben!

                    On Di, 30 Okt 2012, Ben Schmidt wrote:

                    > On 30/10/12 2:40 AM, Jürgen Krämer wrote:
                    > >Hi,
                    > >
                    > >Christian Brabandt wrote:
                    > >>On Mon, October 29, 2012 16:17, Axel wrote:
                    > >>>"s" makes the cursor move to the beginning of the line(!) changing the
                    > >>>character found there (hence the "$" sign). Also, the character "^" is in
                    > >>>fact not displayed.
                    > >>
                    > >>And how do you start Vim? Did you test with vim -u NONE -U NONE -N
                    > >>Does cl work like s (it should, since s is an alias to cl)?
                    > >
                    > >this behavior does not depend on using cl or s. The bug lies in ignoring
                    > >the caret in insert mode and using it later before executing the next
                    > >command. It seems that only every other (dead-letter) caret is inserted
                    > >even if it is followed by a space.
                    >
                    > Caret is not a dead letter for me. So is this bug locale-dependent?

                    It is for a German layout usually. But I can't reproduce it. And
                    possibly also compiler or architecture (32/64bit) dependent.

                    Mit freundlichen Grüßen
                    Christian
                    --
                    Atomkraft, strahlender Glanz ohne Abtrocknen.

                    --
                    You received this message from the "vim_dev" maillist.
                    Do not top-post! Type your reply below the text you are replying to.
                    For more information, visit http://www.vim.org/maillist.php
                  • Jürgen Krämer
                    Hi, ... yesterday, with the example given by Alex I could reproduce it on every try. Today I tried to construct another example, but it worked only randomly --
                    Message 9 of 24 , Oct 29, 2012
                    • 0 Attachment
                      Hi,

                      Christian Brabandt wrote:
                      >
                      > On Di, 30 Okt 2012, Ben Schmidt wrote:
                      >
                      >> On 30/10/12 2:40 AM, Jürgen Krämer wrote:
                      >>> Hi,
                      >>>
                      >>> Christian Brabandt wrote:
                      >>>> On Mon, October 29, 2012 16:17, Axel wrote:
                      >>>>> "s" makes the cursor move to the beginning of the line(!) changing the
                      >>>>> character found there (hence the "$" sign). Also, the character "^" is in
                      >>>>> fact not displayed.
                      >>>>
                      >>>> And how do you start Vim? Did you test with vim -u NONE -U NONE -N
                      >>>> Does cl work like s (it should, since s is an alias to cl)?
                      >>>
                      >>> this behavior does not depend on using cl or s. The bug lies in ignoring
                      >>> the caret in insert mode and using it later before executing the next
                      >>> command. It seems that only every other (dead-letter) caret is inserted
                      >>> even if it is followed by a space.
                      >>
                      >> Caret is not a dead letter for me. So is this bug locale-dependent?
                      >
                      > It is for a German layout usually. But I can't reproduce it. And
                      > possibly also compiler or architecture (32/64bit) dependent.

                      yesterday, with the example given by Alex I could reproduce it on every
                      try. Today I tried to construct another example, but it worked only
                      randomly -- sometimes pressing the caret (or apostrophe or backtick) key
                      followed by one or more presses of the space bar produced the correct
                      character immediately, sometimes it was postponed until the next
                      non-space key was pressed. When it was a letter that could be combined
                      with the respective accent (like "a" or "e") this was done, in other
                      cases like "m" or "," the caret and the letter were inserted separately.

                      I had a look at the source code in gui_w48.c and gui_w32.c, but I could
                      not determine where the dead letter key and the following key are
                      combined into one character and wether this actually has to be done by
                      the application or if it's done by Windows.

                      Regards,
                      Jürgen

                      --
                      Sometimes I think the surest sign that intelligent life exists elsewhere
                      in the universe is that none of it has tried to contact us. (Calvin)

                      --
                      You received this message from the "vim_dev" maillist.
                      Do not top-post! Type your reply below the text you are replying to.
                      For more information, visit http://www.vim.org/maillist.php
                    • Jürgen Krämer
                      Hi, ... additional observation: whenever the caret is not inserted after pressing the space bar and I use the mouse to switch to another program that allows
                      Message 10 of 24 , Oct 29, 2012
                      • 0 Attachment
                        Hi,

                        Jürgen Krämer wrote:
                        >
                        > Christian Brabandt wrote:
                        >>
                        >> On Di, 30 Okt 2012, Ben Schmidt wrote:
                        >>
                        >>> On 30/10/12 2:40 AM, Jürgen Krämer wrote:
                        >>>> Hi,
                        >>>>
                        >>>> Christian Brabandt wrote:
                        >>>>> On Mon, October 29, 2012 16:17, Axel wrote:
                        >>>>>> "s" makes the cursor move to the beginning of the line(!) changing the
                        >>>>>> character found there (hence the "$" sign). Also, the character "^" is in
                        >>>>>> fact not displayed.
                        >>>>>
                        >>>>> And how do you start Vim? Did you test with vim -u NONE -U NONE -N
                        >>>>> Does cl work like s (it should, since s is an alias to cl)?
                        >>>>
                        >>>> this behavior does not depend on using cl or s. The bug lies in ignoring
                        >>>> the caret in insert mode and using it later before executing the next
                        >>>> command. It seems that only every other (dead-letter) caret is inserted
                        >>>> even if it is followed by a space.
                        >>>
                        >>> Caret is not a dead letter for me. So is this bug locale-dependent?
                        >>
                        >> It is for a German layout usually. But I can't reproduce it. And
                        >> possibly also compiler or architecture (32/64bit) dependent.
                        >
                        > yesterday, with the example given by Alex I could reproduce it on every
                        > try. Today I tried to construct another example, but it worked only
                        > randomly -- sometimes pressing the caret (or apostrophe or backtick) key
                        > followed by one or more presses of the space bar produced the correct
                        > character immediately, sometimes it was postponed until the next
                        > non-space key was pressed. When it was a letter that could be combined
                        > with the respective accent (like "a" or "e") this was done, in other
                        > cases like "m" or "," the caret and the letter were inserted separately.
                        >
                        > I had a look at the source code in gui_w48.c and gui_w32.c, but I could
                        > not determine where the dead letter key and the following key are
                        > combined into one character and wether this actually has to be done by
                        > the application or if it's done by Windows.

                        additional observation: whenever the caret is not inserted after pressing
                        the space bar and I use the mouse to switch to another program that allows
                        input, pressing the space bar there results in the caret being inserted.
                        It seems Gvim does not remove the dead letters from the input queue at the
                        correct moment.

                        A similar behavior can be seen if, e.g., in Notepad one presses the caret
                        key, immediately switches to another program by using the mouse, and then
                        presses the space bar: the caret is inserted in the other program, but I
                        guess this is to be expected.

                        jkr

                        --
                        Sometimes I think the surest sign that intelligent life exists elsewhere
                        in the universe is that none of it has tried to contact us. (Calvin)

                        --
                        You received this message from the "vim_dev" maillist.
                        Do not top-post! Type your reply below the text you are replying to.
                        For more information, visit http://www.vim.org/maillist.php
                      • Tony Mechelynck
                        ... [...] ... m or , don t accept a circumflex or acute accent IIUC. OTOH, c g h j s can have a circumflex in Esperanto, and IIRC w is a vowel in Welsh and can
                        Message 11 of 24 , Oct 30, 2012
                        • 0 Attachment
                          On 30/10/12 07:32, Jürgen Krämer wrote:
                          >
                          > Hi,
                          >
                          > Jürgen Krämer wrote:
                          >>
                          >> Christian Brabandt wrote:
                          [...]
                          >>>
                          >>> It is for a German layout usually. But I can't reproduce it. And
                          >>> possibly also compiler or architecture (32/64bit) dependent.
                          >>
                          >> yesterday, with the example given by Alex I could reproduce it on every
                          >> try. Today I tried to construct another example, but it worked only
                          >> randomly -- sometimes pressing the caret (or apostrophe or backtick) key
                          >> followed by one or more presses of the space bar produced the correct
                          >> character immediately, sometimes it was postponed until the next
                          >> non-space key was pressed. When it was a letter that could be combined
                          >> with the respective accent (like "a" or "e") this was done, in other
                          >> cases like "m" or "," the caret and the letter were inserted separately.

                          m or , don't accept a circumflex or acute accent IIUC. OTOH, c g h j s
                          can have a circumflex in Esperanto, and IIRC w is a vowel in Welsh and
                          can accept accents.

                          >>
                          >> I had a look at the source code in gui_w48.c and gui_w32.c, but I could
                          >> not determine where the dead letter key and the following key are
                          >> combined into one character and wether this actually has to be done by
                          >> the application or if it's done by Windows.

                          the application ought to be able to accept characters from the keyboard,
                          e.g. as text input, without even knowing (or caring) whether or not the
                          current keyboard layout includes dead keys.

                          >
                          > additional observation: whenever the caret is not inserted after pressing
                          > the space bar and I use the mouse to switch to another program that allows
                          > input, pressing the space bar there results in the caret being inserted.
                          > It seems Gvim does not remove the dead letters from the input queue at the
                          > correct moment.

                          When gvim does not insert a caret after pressing the dead key then the
                          space bar, what happens if you press the space bar again? (Does the
                          caret appear?) Or if you press some vowel key instead? (Does it insert
                          an accented vowel, e.g. ê if you pressed the e key?) — Another
                          expreiment: if you press the space bar and some letter key in rapid
                          alternation, e.g. x x x x x x x x x x x x x x x x x x x x x , are all
                          letters separated from each other by one space? (Though when I tried,
                          using one hand for each, I noticed that my hand didn't always hit the
                          keyboard in strict alternation.)

                          >
                          > A similar behavior can be seen if, e.g., in Notepad one presses the caret
                          > key, immediately switches to another program by using the mouse, and then
                          > presses the space bar: the caret is inserted in the other program, but I
                          > guess this is to be expected.
                          >
                          > jkr
                          >

                          Sounds like a Windows bug in the keyboard driver to me (not properly
                          combining the dead-key event with the keypress event for the following
                          spacing key, at least in some circumstances, perhaps conditional on a
                          call to some "is there a key waiting?" function not used by Notepad),
                          but I could be wrong.

                          Best regards,
                          Tony.
                          --
                          Forms follow function, and often obliterate it.

                          --
                          You received this message from the "vim_dev" maillist.
                          Do not top-post! Type your reply below the text you are replying to.
                          For more information, visit http://www.vim.org/maillist.php
                        • Jürgen Krämer
                          Hello Tony, ... it s probably more a question of the encoding than of the language which letter will be produced. I just wanted to use a key of which I was
                          Message 12 of 24 , Oct 30, 2012
                          • 0 Attachment
                            Hello Tony,

                            Tony Mechelynck schrieb:
                            > On 30/10/12 07:32, Jürgen Krämer wrote:
                            >>
                            >> Jürgen Krämer wrote:
                            >>>
                            >>> Christian Brabandt wrote:
                            > [...]
                            >>>>
                            >>>> It is for a German layout usually. But I can't reproduce it. And
                            >>>> possibly also compiler or architecture (32/64bit) dependent.
                            >>>
                            >>> yesterday, with the example given by Alex I could reproduce it on every
                            >>> try. Today I tried to construct another example, but it worked only
                            >>> randomly -- sometimes pressing the caret (or apostrophe or backtick) key
                            >>> followed by one or more presses of the space bar produced the correct
                            >>> character immediately, sometimes it was postponed until the next
                            >>> non-space key was pressed. When it was a letter that could be combined
                            >>> with the respective accent (like "a" or "e") this was done, in other
                            >>> cases like "m" or "," the caret and the letter were inserted separately.
                            >
                            > m or , don't accept a circumflex or acute accent IIUC. OTOH, c g h j s
                            > can have a circumflex in Esperanto, and IIRC w is a vowel in Welsh and
                            > can accept accents.

                            it's probably more a question of the encoding than of the language which
                            letter will be produced. I just wanted to use a key of which I was sure
                            that when pressed after the dead letter caret would produce the caret
                            and the corresponding character. "m" and "," are one of those.

                            The caret key followed by space bar should always produce a caret
                            character on its own, but sometimes it does not.

                            BTW, at least Gvim on Windows with enc=utf-8 and fenc=utf-8 does not
                            seem to produce one of the accented letters you listed, although
                            pre-composed characters exist for them in the Unicode standard.

                            >>> I had a look at the source code in gui_w48.c and gui_w32.c, but I could
                            >>> not determine where the dead letter key and the following key are
                            >>> combined into one character and wether this actually has to be done by
                            >>> the application or if it's done by Windows.
                            >
                            > the application ought to be able to accept characters from the keyboard,
                            > e.g. as text input, without even knowing (or caring) whether or not the
                            > current keyboard layout includes dead keys.

                            On Windows there are two messages (WM_DEADCHAR and WM_SYSDEADCHAR) which
                            are sent to a program when Windows has processed dead letter keys. Gvim
                            reacts to both of them by setting a flag in int _OnDeadChar() function.
                            I don't know whether this behavior is superfluous, incomplete or wrong.

                            >> additional observation: whenever the caret is not inserted after pressing
                            >> the space bar and I use the mouse to switch to another program that allows
                            >> input, pressing the space bar there results in the caret being inserted.
                            >> It seems Gvim does not remove the dead letters from the input queue at the
                            >> correct moment.
                            >
                            > When gvim does not insert a caret after pressing the dead key then the
                            > space bar, what happens if you press the space bar again? (Does the
                            > caret appear?)

                            No, it's only after a non-space key that the caret appears -- either on
                            its own if it's a consonant or as part of an accented letter if it's a
                            vowel.

                            > Or if you press some vowel key instead? (Does it insert
                            > an accented vowel, e.g. ê if you pressed the e key?) — Another
                            > expreiment: if you press the space bar and some letter key in rapid
                            > alternation, e.g. x x x x x x x x x x x x x x x x x x x x x , are all
                            > letters separated from each other by one space? (Though when I tried,
                            > using one hand for each, I noticed that my hand didn't always hit the
                            > keyboard in strict alternation.)

                            Alternating between space bar and some letter key works correctly. It's
                            only the dead letter keys in combination with space bar that don't
                            always produce the spacing, non-combining character for the accent. And
                            when this happens (e.g., caret followed by space produces neither of
                            both), a second caret produces two carets.

                            >> A similar behavior can be seen if, e.g., in Notepad one presses the caret
                            >> key, immediately switches to another program by using the mouse, and then
                            >> presses the space bar: the caret is inserted in the other program, but I
                            >> guess this is to be expected.
                            >
                            > Sounds like a Windows bug in the keyboard driver to me (not properly
                            > combining the dead-key event with the keypress event for the following
                            > spacing key, at least in some circumstances, perhaps conditional on a
                            > call to some "is there a key waiting?" function not used by Notepad),
                            > but I could be wrong.

                            I fear it's a bug in Gvim. Windows has messages that can be handled by
                            the application. Some hold the combined character in their parameters
                            (like WM_CHAR) and some -- WM_KEYDOWN and WM_KEYUP -- hold information
                            on which physical key was pressed. Gvim handles these messages inside
                            its process_message() function and I guess there can go something wrong
                            with setting and resetting the dead_key variable.

                            Regards,
                            Jürgen

                            --
                            Sometimes I think the surest sign that intelligent life exists elsewhere
                            in the universe is that none of it has tried to contact us. (Calvin)

                            --
                            You received this message from the "vim_dev" maillist.
                            Do not top-post! Type your reply below the text you are replying to.
                            For more information, visit http://www.vim.org/maillist.php
                          • Axel
                            I think the fact that this behavior is not evident in the original (release) version of gvim (7.3.046) demonstrates that this is a bug. -- You received this
                            Message 13 of 24 , Oct 31, 2012
                            • 0 Attachment
                              I think the fact that this behavior is not evident in the original (release) version of gvim (7.3.046) demonstrates that this is a bug.

                              --
                              You received this message from the "vim_dev" maillist.
                              Do not top-post! Type your reply below the text you are replying to.
                              For more information, visit http://www.vim.org/maillist.php
                            • Tony Mechelynck
                              ... On my system I get them: ĉ ĝ ĥ ĵ ŝ ŵ (using the dead-^ right of the P on my Belgian AZERTY keyboard). The superiority of X11 keyboard drivers over
                              Message 14 of 24 , Oct 31, 2012
                              • 0 Attachment
                                On 31/10/12 07:27, Jürgen Krämer wrote:
                                >
                                > Hello Tony,
                                >
                                > Tony Mechelynck schrieb:
                                >> On 30/10/12 07:32, Jürgen Krämer wrote:
                                >>>
                                >>> Jürgen Krämer wrote:
                                >>>>
                                >>>> Christian Brabandt wrote:
                                >> [...]
                                >>>>>
                                >>>>> It is for a German layout usually. But I can't reproduce it. And
                                >>>>> possibly also compiler or architecture (32/64bit) dependent.
                                >>>>
                                >>>> yesterday, with the example given by Alex I could reproduce it on every
                                >>>> try. Today I tried to construct another example, but it worked only
                                >>>> randomly -- sometimes pressing the caret (or apostrophe or backtick) key
                                >>>> followed by one or more presses of the space bar produced the correct
                                >>>> character immediately, sometimes it was postponed until the next
                                >>>> non-space key was pressed. When it was a letter that could be combined
                                >>>> with the respective accent (like "a" or "e") this was done, in other
                                >>>> cases like "m" or "," the caret and the letter were inserted separately.
                                >>
                                >> m or , don't accept a circumflex or acute accent IIUC. OTOH, c g h j s
                                >> can have a circumflex in Esperanto, and IIRC w is a vowel in Welsh and
                                >> can accept accents.
                                >
                                > it's probably more a question of the encoding than of the language which
                                > letter will be produced. I just wanted to use a key of which I was sure
                                > that when pressed after the dead letter caret would produce the caret
                                > and the corresponding character. "m" and "," are one of those.
                                >
                                > The caret key followed by space bar should always produce a caret
                                > character on its own, but sometimes it does not.
                                >
                                > BTW, at least Gvim on Windows with enc=utf-8 and fenc=utf-8 does not
                                > seem to produce one of the accented letters you listed, although
                                > pre-composed characters exist for them in the Unicode standard.

                                On my system I get them: ĉ ĝ ĥ ĵ ŝ ŵ (using the dead-^ right of the P on
                                my Belgian AZERTY keyboard). The superiority of X11 keyboard drivers
                                over those of Windows, I suppose. See also
                                http://users.skynet.be/antoine.mechelynck/other/keybbe.htm

                                >
                                >>>> I had a look at the source code in gui_w48.c and gui_w32.c, but I could
                                >>>> not determine where the dead letter key and the following key are
                                >>>> combined into one character and wether this actually has to be done by
                                >>>> the application or if it's done by Windows.
                                >>
                                >> the application ought to be able to accept characters from the keyboard,
                                >> e.g. as text input, without even knowing (or caring) whether or not the
                                >> current keyboard layout includes dead keys.
                                >
                                > On Windows there are two messages (WM_DEADCHAR and WM_SYSDEADCHAR) which
                                > are sent to a program when Windows has processed dead letter keys. Gvim
                                > reacts to both of them by setting a flag in int _OnDeadChar() function.
                                > I don't know whether this behavior is superfluous, incomplete or wrong.

                                Aha! Well, Vim (or, at least, gvim) wants to be able to track wild key
                                combinations like Shift-Alt-F12 or Ctrl-Alt-PgUp, but dead keys ought to
                                be a different matter. However, see os_win32.txt lines 212 sqq. I
                                thought it didn't apply to Win-XP and later, but who knows?

                                >
                                >>> additional observation: whenever the caret is not inserted after pressing
                                >>> the space bar and I use the mouse to switch to another program that allows
                                >>> input, pressing the space bar there results in the caret being inserted.
                                >>> It seems Gvim does not remove the dead letters from the input queue at the
                                >>> correct moment.
                                >>
                                >> When gvim does not insert a caret after pressing the dead key then the
                                >> space bar, what happens if you press the space bar again? (Does the
                                >> caret appear?)
                                >
                                > No, it's only after a non-space key that the caret appears -- either on
                                > its own if it's a consonant or as part of an accented letter if it's a
                                > vowel.
                                >
                                >> Or if you press some vowel key instead? (Does it insert
                                >> an accented vowel, e.g. ê if you pressed the e key?) — Another
                                >> expreiment: if you press the space bar and some letter key in rapid
                                >> alternation, e.g. x x x x x x x x x x x x x x x x x x x x x , are all
                                >> letters separated from each other by one space? (Though when I tried,
                                >> using one hand for each, I noticed that my hand didn't always hit the
                                >> keyboard in strict alternation.)
                                >
                                > Alternating between space bar and some letter key works correctly. It's
                                > only the dead letter keys in combination with space bar that don't
                                > always produce the spacing, non-combining character for the accent. And
                                > when this happens (e.g., caret followed by space produces neither of
                                > both), a second caret produces two carets.

                                Hm.

                                >
                                >>> A similar behavior can be seen if, e.g., in Notepad one presses the caret
                                >>> key, immediately switches to another program by using the mouse, and then
                                >>> presses the space bar: the caret is inserted in the other program, but I
                                >>> guess this is to be expected.
                                >>
                                >> Sounds like a Windows bug in the keyboard driver to me (not properly
                                >> combining the dead-key event with the keypress event for the following
                                >> spacing key, at least in some circumstances, perhaps conditional on a
                                >> call to some "is there a key waiting?" function not used by Notepad),
                                >> but I could be wrong.
                                >
                                > I fear it's a bug in Gvim. Windows has messages that can be handled by
                                > the application. Some hold the combined character in their parameters
                                > (like WM_CHAR) and some -- WM_KEYDOWN and WM_KEYUP -- hold information
                                > on which physical key was pressed. Gvim handles these messages inside
                                > its process_message() function and I guess there can go something wrong
                                > with setting and resetting the dead_key variable.
                                >
                                > Regards,
                                > Jürgen
                                >
                                Let's hope someone reads this thread with better knowledge of this stuff
                                than I have.

                                Best regards,
                                Tony.
                                --
                                If Helen Keller is alone in a forest and falls, does she make a sound?


                                --
                                You received this message from the "vim_dev" maillist.
                                Do not top-post! Type your reply below the text you are replying to.
                                For more information, visit http://www.vim.org/maillist.php
                              • Tony Mechelynck
                                ... It should be possible (but a lot of work) to determine a last good and a first bad version by binary search (dichotomy) and experiment, maybe with the
                                Message 15 of 24 , Oct 31, 2012
                                • 0 Attachment
                                  On 31/10/12 08:04, Axel wrote:
                                  > I think the fact that this behavior is not evident in the original (release) version of gvim (7.3.046) demonstrates that this is a bug.
                                  >

                                  It should be possible (but a lot of work) to determine a "last good" and
                                  a "first bad" version by binary search (dichotomy) and experiment, maybe
                                  with the help of with the W32 versions of gvim at
                                  http://sourceforge.net/projects/cream/files/Vim/ — where the last
                                  version seems mislabeled: it says 6.3.709 on the list of versions but it
                                  offers gvim-7-3-709.exe and gvim-7-3-709_release-notes.txt for downloading.

                                  The procedure is as follows: You already have a "known good" (7.3.046)
                                  and a "known bad" (7.3.712). Try a version somewhere between them. If it
                                  is "good" you have a later "known good"; if it is bad, an earlier "known
                                  bad", so in either case you've narrowed the error range a bit. Try again
                                  until you find the "last good" and the "first bad" and they are as close
                                  as possible to each other. This would help find when the regression
                                  crept in.


                                  Best regards,
                                  Tony.
                                  --
                                  Life is a gift, living is an art. (Bram Moolenaar)

                                  --
                                  You received this message from the "vim_dev" maillist.
                                  Do not top-post! Type your reply below the text you are replying to.
                                  For more information, visit http://www.vim.org/maillist.php
                                • Axel
                                  Following Tony s approach I tried to verify the behavior with Cream s 7.3.709 but I couldn t; gvim.exe behaves correctly. The Cream version however is a 32-bit
                                  Message 16 of 24 , Oct 31, 2012
                                  • 0 Attachment
                                    Following Tony's approach I tried to verify the behavior with Cream's 7.3.709 but I couldn't; gvim.exe behaves correctly. The Cream version however is a 32-bit application, where mine is a 64-bit version compiled with MinGW.

                                    @Jürgen

                                    For which platform did you compile/download your version of gvim.exe?

                                    --
                                    You received this message from the "vim_dev" maillist.
                                    Do not top-post! Type your reply below the text you are replying to.
                                    For more information, visit http://www.vim.org/maillist.php
                                  • Ben Fritz
                                    ... Since you re compiling yourself, Mercurial actually automates Tony s approach with it s bisect feature. I assume you re using Mercurial to get the Vim
                                    Message 17 of 24 , Oct 31, 2012
                                    • 0 Attachment
                                      On Wednesday, October 31, 2012 5:48:51 AM UTC-5, Axel wrote:
                                      > Following Tony's approach I tried to verify the behavior with Cream's 7.3.709 but I couldn't; gvim.exe behaves correctly. The Cream version however is a 32-bit application, where mine is a 64-bit version compiled with MinGW.
                                      >
                                      >

                                      Since you're compiling yourself, Mercurial actually automates Tony's approach with it's "bisect" feature. I assume you're using Mercurial to get the Vim source?

                                      See:

                                      http://www.selenic.com/mercurial/hg.1.html#bisect
                                      http://tortoisehg.bitbucket.org/hgbook/1.6/finding-and-fixing-mistakes.html#sec:undo:bisect

                                      (I expected the second link to show how to do it graphically in TortoiseHg but I was mistaken; it looks to be a good read whether you use the command-line or Tortoise tools).

                                      --
                                      You received this message from the "vim_dev" maillist.
                                      Do not top-post! Type your reply below the text you are replying to.
                                      For more information, visit http://www.vim.org/maillist.php
                                    • Tony Mechelynck
                                      ... Yes, and, of course, hg help bisect will let you RTFM even if you aren t on Linux ;-) Best regards, Tony. -- Around computers it is difficult to find the
                                      Message 18 of 24 , Oct 31, 2012
                                      • 0 Attachment
                                        On 31/10/12 15:38, Ben Fritz wrote:
                                        > On Wednesday, October 31, 2012 5:48:51 AM UTC-5, Axel wrote:
                                        >> Following Tony's approach I tried to verify the behavior with Cream's 7.3.709 but I couldn't; gvim.exe behaves correctly. The Cream version however is a 32-bit application, where mine is a 64-bit version compiled with MinGW.
                                        >>
                                        >>
                                        >
                                        > Since you're compiling yourself, Mercurial actually automates Tony's approach with it's "bisect" feature. I assume you're using Mercurial to get the Vim source?
                                        >
                                        > See:
                                        >
                                        > http://www.selenic.com/mercurial/hg.1.html#bisect
                                        > http://tortoisehg.bitbucket.org/hgbook/1.6/finding-and-fixing-mistakes.html#sec:undo:bisect
                                        >
                                        > (I expected the second link to show how to do it graphically in TortoiseHg but I was mistaken; it looks to be a good read whether you use the command-line or Tortoise tools).
                                        >

                                        Yes, and, of course,

                                        hg help bisect

                                        will let you RTFM even if you aren't on Linux ;-)


                                        Best regards,
                                        Tony.
                                        --
                                        Around computers it is difficult to find the correct unit of time to
                                        measure progress. Some cathedrals took a century to complete. Can you
                                        imagine the grandeur and scope of a program that would take as long?
                                        -- Epigrams in Programming, ACM SIGPLAN Sept. 1982

                                        --
                                        You received this message from the "vim_dev" maillist.
                                        Do not top-post! Type your reply below the text you are replying to.
                                        For more information, visit http://www.vim.org/maillist.php
                                      • Steve Hall
                                        On Wed, Oct 31, 2012 at 6:31 AM, Tony Mechelynck ... Whoops, fixed this folder name for the record. -- Steve Hall [ digitect dancingpaper com ] -- You
                                        Message 19 of 24 , Oct 31, 2012
                                        • 0 Attachment
                                          On Wed, Oct 31, 2012 at 6:31 AM, Tony Mechelynck
                                          <antoine.mechelynck@...> wrote:
                                          >
                                          > ....the W32 versions of gvim at
                                          > http://sourceforge.net/projects/cream/files/Vim/ — where the last
                                          > version seems mislabeled: it says 6.3.709 on the list of versions
                                          > but it offers gvim-7-3-709.exe and gvim-7-3-709_release-notes.txt
                                          > for downloading.

                                          Whoops, fixed this folder name for the record.

                                          --
                                          Steve Hall [ digitect dancingpaper com ]

                                          --
                                          You received this message from the "vim_dev" maillist.
                                          Do not top-post! Type your reply below the text you are replying to.
                                          For more information, visit http://www.vim.org/maillist.php
                                        • Christian Brabandt
                                          ... I still couldn t reproduce the issue with VisualStudio 2010 Express. But it only compiled a 32bit version and I couldn t convince Windows to let me install
                                          Message 20 of 24 , Nov 1, 2012
                                          • 0 Attachment
                                            On Mo, 29 Okt 2012, Jürgen Krämer wrote:

                                            > I can confirm this behavior on Vim 7.3.1-712 on Windows 7, compiled with
                                            > MS-C 16.0.40219.1 (i.e., Visual Studio 2008), although it only seems to
                                            > happen of every other input of "^".
                                            >
                                            > The caret is a dead-letter key and is ignored although a space is pressed
                                            > afterwards. It seems to be kept in the input buffer, though, and is
                                            > finally used when "s" is pressed, leading to a movement to the start of
                                            > line and starting insert mode at the wrong position.
                                            >
                                            > The dollar sign you see in Alex' example is the one displayed at the end
                                            > of the changed text if "$" is included in 'cpo'.

                                            I still couldn't reproduce the issue with VisualStudio 2010 Express. But
                                            it only compiled a 32bit version and I couldn't convince Windows to let
                                            me install the required 64bit utilities to built a 64bit version.

                                            regards,
                                            Christian

                                            --
                                            You received this message from the "vim_dev" maillist.
                                            Do not top-post! Type your reply below the text you are replying to.
                                            For more information, visit http://www.vim.org/maillist.php
                                          Your message has been successfully submitted and would be delivered to recipients shortly.