-
Notifications
You must be signed in to change notification settings - Fork 61
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Win32 build #520
Win32 build #520
Conversation
test_file_lines now counts the lines of the oldest procedure in src/pro hash, list are still incomplete so some tests are skipped if "git" is found in !gdl.release. MYMESS exchanged out for ERRORS_ADD Greg Jung (aka maynardGK) gvjung@gmail.com
These are the result of a lot of experimentation - the scripting is very fragile. Nevertheless some flexibilty in the builds is allowed with small manipulation of the appveyor.yml file. For instance, a value for $env:platform value of "mingw64630x8664" will cause gdl to be built using the scripts/mingw-bldgdl.bat script and mingw32-make. So, after using $env:platform to establish other variables, the default build (using MSYS2) is forced by setting $env:platform=""
…ilds. In MINGW32, the functions that were found in PCRE for regular expressions are now derived from libsystre. In Linux,there is no "systre" lib and so the routine will find, as usual, PCRE. Similarly for the early version readline5 of win32libs, systre will not be found and pcre can satisfy the demand. The compilation of datatypes.cpp created a "file too big" error when compiling for 64-bits, and so CMakeLists.txt includes an option, add_compile_options(-Wa,-mbig-obj), for the WIN32 case.
test_obj_new and test_typename were failing in WIN32 because the object example they used was too large for GDL/WIN32 - a different error than what the tests are supposed to test.
Codecov Report
@@ Coverage Diff @@
## master #520 +/- ##
==========================================
+ Coverage 43.79% 43.79% +<.01%
==========================================
Files 295 295
Lines 95839 95846 +7
==========================================
+ Hits 41969 41979 +10
+ Misses 53870 53867 -3
Continue to review full report at Codecov.
|
…n systems. Is has been found that OSX/clang builds, in the travis builds with full dependencies, fail when a wxInitialize is called. I wouldn't expect wxWidgets to work at all on Darwin systems, or maybe there is simply a flaw in the Travis implementation.
isRange is the commonly used name for an array that gets passed in indexing, that indicates whether an index value or an index range is denoted. It is a one-dimensional GDL variable that has up to now been incorrectly formed.
also ::GetDecomposed(),::Decomposed(), ::GetVisualDepth() as for devicex.
@GillesDuvert Could you please merge this |
The appveyor tests on the windows' builds indicate segfaults and errors, but this is not the behavior On a newly built Win10 (32-bit) system (a copy of the testsuite/ folder, from the git distribution, is needed as a subdirectory of gdllib/ in the artifact, to run the procedures):
|
This PR collects results from #512 and #514 in the first few commits.
Then, a few tests are adapted for differences in operating systems.
In file.cpp, File_search routines used in Win32 case (glob is not available) is tweaked, passing test_file_search.
Lastly,
widget initialization routine wxInitialize() is called during gdl initialization, however for darwin
systems it is not called at all, as this seems to make these GDL sick (travis builds xxx.5, xxx.10)!
To test this on an OSX computer, the routine dialog_pickfile will still call this.