Re: [Lazarus] Default ico handling and recent debugger errors

2015-03-23 Thread JuuS

A bit more information on this topic if anyone is interested.

The debugger error occurred on three machines (however all three are
Kubuntu 14.04's).

I tried a test with a debug Lazarus IDE itself on one of the other
machines that still has the unfixed debugger unit. Making sure the
Lazarus IDE was shut down I renamed lazarus.ico to lazarus.xxxico.

I then opened lazarus (the missing ico file exception is silent) and
then tried to run lazarus IDE as a slave from this master. And the
debugger error comes up immediately and the IDE is broken.

Again this won't happen if the new debugger code has been added to the
system.

It just seems to me this may be of interest in that it caused such a
fatal error in debugging and breaks the program. Maybe it has other
unforeseen effects?

Can anyone reproduce or is is just my Universe?


On 03/21/2015 02:32 PM, JuuS wrote:
 
 While testing RC1 and 2 I followed the suggestion to open a published
 project and see how the release candidates were doing with them.
 
 Upon doing this I got an immediate debugger error, which has been traced
 down now (as of last night) with the help of Martin and is fixed.
 
 The problems origins were (and still are to some extent)...weird.
 
 Before a debugger solution was found I could stop the error by simply
 removing the default icon line from the published project's LPI file,
 then no debugger error.
 
 During all the research I could see that while the LPI file had a
 default icon line, the publishing process did not copy the ico file to
 the published diretory.
 
 So there was always a background silent error due to the fact that the
 icon file didn't exist, the silent exception (seen when debugging
 Lazarus) was on trying to open this nonexisting file.
 
 How this all caused the debugger error is, as I have said, still not
 entirely understood.
 
 So it just brings to my mind that even though this error has no effect
 in most cases it may be best to either copy the ico file on publishing
 the project, or not writing a default icon line in LPI, or checking
 that the ico file actually does exist and recreating it if not; just in
 case it has other effects that will cause other weird errors down the
 line.
 


--
___
Lazarus mailing list
Lazarus@lists.lazarus.freepascal.org
http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus


[Lazarus] Default ico handling and recent debugger errors

2015-03-21 Thread JuuS

While testing RC1 and 2 I followed the suggestion to open a published
project and see how the release candidates were doing with them.

Upon doing this I got an immediate debugger error, which has been traced
down now (as of last night) with the help of Martin and is fixed.

The problems origins were (and still are to some extent)...weird.

Before a debugger solution was found I could stop the error by simply
removing the default icon line from the published project's LPI file,
then no debugger error.

During all the research I could see that while the LPI file had a
default icon line, the publishing process did not copy the ico file to
the published diretory.

So there was always a background silent error due to the fact that the
icon file didn't exist, the silent exception (seen when debugging
Lazarus) was on trying to open this nonexisting file.

How this all caused the debugger error is, as I have said, still not
entirely understood.

So it just brings to my mind that even though this error has no effect
in most cases it may be best to either copy the ico file on publishing
the project, or not writing a default icon line in LPI, or checking
that the ico file actually does exist and recreating it if not; just in
case it has other effects that will cause other weird errors down the
line.




On 03/21/2015 02:04 PM, JuuS wrote:
 What a coincidence!~
 
 I just wrote a message to board about this...it is ibus.
 
 go to terminal: type: ibus exit
 
 Your problems will go away...
 
 
 
 On 03/21/2015 11:26 AM, Michael Van Canneyt wrote:

 Gentlemen,

 I don't know how best to explain this problem.

 Platform: Linux, 64 bit PC. GTK

 Today, I started a recent lazarus 1.5 on a new PC and started working in
 it. After some time I noticed the m key (on an azerty keyboard) no
 longer worked in the source editor.
 Prior to that I had not touched the M key.

 In synedit, The M behaves as a 'ctrl-m' key: a CR (^13) is inserted !
 I know the synedit is a bit special. So I decided to test the key in
 another edit control.

 Pressing shift-M works as expected and inserts a capital M.

 When typing in the project options, title option edit control, I got
 double m characters.

 So, I then updated lazarus from SVN to todays version and recompiled.
 The problem persisted.
 Then, when typing in any edit box, now I get double letters (a - aa etc)

 as soon as I touch the m key, all input just stops, I can no longer edit
 anything.
 Restarting lazarus allows me to type characters, until I touch the M key
 outside of synedit.

 I had several older lazarus binaries on this machine, so I tried that:
 there the 'm' key works fine.

 The strange thing is, if I go to 'keyboard mappings' 'Find key
 combination' and press 'Grab key' then the key is displayed fine.

 The m key works as expected in all other programs:
 I am typing this mail on the same machine, and the M key works just fine.

 All this makes me pretty confident it is a Lazarus issue, probably
 trying to handle the
 CTRL-M (CR) key in some special (global) way.

 Suggestions for fixing this ?

 Michael.

 -- 
 ___
 Lazarus mailing list
 Lazarus@lists.lazarus.freepascal.org
 http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
 
 --
 ___
 Lazarus mailing list
 Lazarus@lists.lazarus.freepascal.org
 http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
 

--
___
Lazarus mailing list
Lazarus@lists.lazarus.freepascal.org
http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus