View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0036160||Lazarus||Debugger||public||2019-10-10 11:33||2019-10-11 13:54|
|Reporter||CudaText man||Assigned To||Martin Friebe|
|Product Version||2.1 (SVN)|
|Summary||0036160: Debugger shows bad IDE tooltips for widechar|
|Description||run this app in IDE. place breakpoint on /if/ line. during the breakpoint stop, place mouse cursor over 'ch' variable.|
IDE shows bad tooltip for first WideChars from the example string.
ie negative numbers like:
procedure TForm1.FormCreate(Sender: TObject);
for i:= 1 to length(s) do
if ch=#0 then begin end;
|Tags||No tags attached.|
|Fixed in Revision|
Version of fpc and gdb?
I get (3.0.4 / 7.7 and 8.2)
s = 'test'
(s)^ = 61600 #61600
tested wrong var, but same for "ch"
Free Pascal Compiler version 3.3.1-r41371 [2019/02/18] for i386
GNU gdb (GDB) 7.2
Please verify the below, and reopen the issue if your case does not match the findings. (Otherwise this is a "wont fix")
I tested with the given setup and the issue only occurred if the debug info type was set to "-gs stabs" or "automatic".
STABS is very old, and has many disadvantages. It is generally recommended to use dwarf instead ("dwarf with sets" with gdb). From my tests issues with stabs will further increase when the IDE will move to more modern (8.x) versions of GDB.
For that reason stabs issues will in most cases not be addressed any more.
About debug info "automatic":
This lets FPC decide, and apparently for win32, fpc still defaults to stabs. (If that is an issue please report to fpc).
It is indeed a problem that the default is debug info (on win32) is causing such issues.
And yes it is arguable that the IDE should have its own defaults.
That is a question of man hours available though. (We will probably have fpdebug before such an ide-default. And fpdebug enforces dwarf).
But of course contributions towards a separate ide-default are welcome.
Also if the issue (negative values with stabs) can be fixed with a small or reasonable small patch (such as possible side effects can be judged without too much effort), and such a patch would be provided, then it can of course be accepted.
Further more, but not related to this:
The 32bit IDE for windows currently ships with gdb 7.7.1
|2019-10-10 11:33||CudaText man||New Issue|
|2019-10-10 11:33||CudaText man||Status||new => assigned|
|2019-10-10 11:33||CudaText man||Assigned To||=> Martin Friebe|
|2019-10-10 13:17||Martin Friebe||Status||assigned => feedback|
|2019-10-10 13:17||Martin Friebe||LazTarget||=> -|
|2019-10-10 13:17||Martin Friebe||Note Added: 0118461|
|2019-10-10 13:19||Martin Friebe||Note Edited: 0118461||View Revisions|
|2019-10-10 14:50||Martin Friebe||Note Edited: 0118461||View Revisions|
|2019-10-10 17:48||CudaText man||Note Added: 0118468|
|2019-10-10 17:48||CudaText man||Status||feedback => assigned|
|2019-10-10 20:51||Martin Friebe||Status||assigned => resolved|
|2019-10-10 20:51||Martin Friebe||Resolution||open => won't fix|
|2019-10-10 20:51||Martin Friebe||Note Added: 0118479|
|2019-10-11 13:54||CudaText man||Status||resolved => closed|