Károly Balogh
|
9693b37c1a
m68k: actually,overwrite the resflags with F_E, because only that would be valid after the FSxx/TST combination
|
10 gadi atpakaļ |
Károly Balogh
|
c062e55aa2
m68k: after a compare on the FPU, move the condition flags back to the CPU. this should make floating point compare actually working
|
10 gadi atpakaļ |
Tomas Hajny
|
46bfe5c1dd
* set ShortDateFormat explicitly so that StrToDate behaviour can be tested easily regardless of current locale
|
10 gadi atpakaļ |
Tomas Hajny
|
c5c2ac13e2
* TODO updated according to the current status of OS/2 RTL
|
10 gadi atpakaļ |
Tomas Hajny
|
41312d44b4
* set DecimalSeparator explicitly so that Format output can be tested easily regardless of current locale
|
10 gadi atpakaļ |
Tomas Hajny
|
7e15bf6c2d
* ensure passing valid stacksize to emxbind
|
10 gadi atpakaļ |
Tomas Hajny
|
e1b37dcc46
* the test made compilable also on other platforms than only Unix and Windows
|
10 gadi atpakaļ |
Tomas Hajny
|
90aeabb439
* skip this test under OS/2, because it expects that replacement used for characters not supported in the current codepage must match between compile time ('?' used by FPC) and run-time (ASCII 127 used by OS/2 native Unicode routines)
|
10 gadi atpakaļ |
Tomas Hajny
|
fbb6776085
* this test results in a SIGSEGV under OS/2, but I consider it a correct behaviour if a nil pointer for passing a nil as a buffer to BlockWrite
|
10 gadi atpakaļ |
Tomas Hajny
|
d2ae1f0ab2
* skip emx too
|
10 gadi atpakaļ |
Tomas Hajny
|
60c5f29b5b
* add path setting for some previously missing platforms (OS/2, MSDOS)
|
10 gadi atpakaļ |
marcus
|
141878978e
AmiCommon: Missing UnLockDosList, DiskFree Bugfix
|
10 gadi atpakaļ |
florian
|
3ed1a75fcd
* write ok on exit for easier testing
|
10 gadi atpakaļ |
sergei
|
d93d36e6d3
* Win64: Use separate implementation of CaptureBacktrace that calls RtlCaptureStackBackTrace. This way it does not require non-optimized code to work correctly.
|
10 gadi atpakaļ |
sergei
|
438c433c50
* CaptureBacktrace: fixed off-by-one error. Caller's context must be captured if skipframes=0.
|
10 gadi atpakaļ |
Tomas Hajny
|
2e2ca8684d
* add resolution of DLL name under OS/2
|
10 gadi atpakaļ |
Tomas Hajny
|
95c0092a54
* skip test if CPU does not support SSE2 (similarly to test/tshuffle1.pp)
|
10 gadi atpakaļ |
svenbarth
|
a7085ea34e
Added test for Mantis #25044 which has been fixed in revision 29537.
|
10 gadi atpakaļ |
Jonas Maebe
|
071282b286
* add methods of objcprotocols to the list of visible "helper" methods when
|
10 gadi atpakaļ |
svenbarth
|
04534d5dcf
Fix for Mantis #27424.
|
10 gadi atpakaļ |
marco
|
8e2f8d7f3c
* for wince use getprocessaddresa, mantis #27450
|
10 gadi atpakaļ |
marco
|
6e23f3fc97
* patch from mantis #27453 by Dirk Fellenberg, update fontsize to freetype
|
10 gadi atpakaļ |
michael
|
890b9d37e4
* Patch from Laco: change PChar -> PAnsiChar where appropriate
|
10 gadi atpakaļ |
michael
|
c3d427210d
* Patch from Laco to fix bug ID #27425
|
10 gadi atpakaļ |
lacak
|
0b545308d7
fcl-db: base: when master dataset is closed leave detail dataset intact. Later if master is reopened relationship will work again. Delphi compatible behavior. Bug #26424 (Grandfather-Father-Son relationship)
|
10 gadi atpakaļ |
sergei
|
896c1ff6d6
* Win64 SEH: don't call RunError in exception handler, because it always prints backtrace from caller's context. Instead, print the correct backtrace explicitly.
|
10 gadi atpakaļ |
Tomas Hajny
|
cd166e6ff6
* modified according to outcomes of discussion with Jonas
|
10 gadi atpakaļ |
Tomas Hajny
|
bd719052fa
* test updated according to results of discussion with Jonas to work properly also on platforms not using UTF-8 as their default system codepage (and be safe to ignore characters possibly unsupported with current system codepage)
|
10 gadi atpakaļ |
Tomas Hajny
|
08a3c45aa3
* test fixed because the original version was wrong (as confirmed by Jonas)
|
10 gadi atpakaļ |
Tomas Hajny
|
da21fe94fa
* this test can never work correctly on operating systems not providing Unicode support at filesystem API level unless the current codepage happens to support all the characters used within the test (which is unlikely at least)
|
10 gadi atpakaļ |