gfortran TDM 5.1 problem

[gd_on]

I tried your new 17.12 C::B provided with gfortran (codeblocks-17.12mingw_fortran-setup.exe).
I think you should not distribute this gfortran version (TDM 5.1.0) because it is bugged (no such problem with TDM gcc 5.1.0).
This is not new. Version TDM 4.9.2 had not this bug. It has been corrected in official 5.2 version, and it works also with recent official versions. Its OK too with 6.4 reckless version (except a problem with an empty drive, but it's an other problem already signaled somewhere in the forum).
See the gfortran test case in attachment.
gd_on

[BentL]

It has been corrected in official 5.2 version, and it works also with recent official versions.

According to this bug report:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66861

this bug was introduced in version 5.1 and corrected in version 5.2.

The bug is in the version 5.1.0 of the file "libgfortran-3.dll" located in the default installation folder:

C:\Program Files (x86)\CodeBlocks\MinGW\bin

A newer version 5.3.0 of this file can be found in file "libgfortran-5.3.0-3-mingw32-dll-3.tar.xz" at:

https://sourceforge.net/projects/mingw/files/MinGW/Base/gcc/Version5/gcc-5.3.0-3/

This file can be opened in e.g. WinRar and version 5.3.0 of the file "libgfortran-3.dll" can be extracted. By replacing the existing version 5.1.0 file with the version 5.3.0 file in the default installation folder mentioned above, reading from files works again.

The gfortran test case in the attachment in the post above works now.

[gd_on]

cool.
nevertheless, mixing 5.1 and 5.3 gfortran versions may cause other problems (but may not  : ). I'll try that.
gd_on

[karaul]

does not work in my case too

[grem]

What about updating/replacing "codeblocks-17.12mingw_fortran-setup.exe" with updated TDM MinGW librariy as current binary contains such serious problem?

What will the replacement of TDM in future? Does it planning to replace it with MinGW64 as currently as I see the TDM development is frozen?

[stahta01]

What about updating/replacing "codeblocks-17.12mingw_fortran-setup.exe" with updated TDM MinGW librariy as current binary contains such serious problem?

What will the replacement of TDM in future? Does it planning to replace it with MinGW64 as currently as I see the TDM development is frozen?

I think we should remove the "codeblocks-17.12mingw_fortran-setup.exe" from the download site.

[gd_on]

I think we should remove the "codeblocks-17.12mingw_fortran-setup.exe" from the download site.

I think a better solution is to propose a recent MinGW version : may be 7.3 (or 8.1, but may be too new !). Personnaly, I have projects mixing C, C++ and fortran codes, in 32 and/or 64 bits. So, I need an homogeneous solution.
What is your feeling devs ?
What is the best solution ?
Which one to choose between pure 32 bits solution, or 64 bits ones ?
If I look at the number of downloads, it looks that for 64 bit, a good choice is seh, posix and for pure 32 bits it's dwarf, posix.
If for 64 bits, it's looks compatible with TDM choices, for pure 32 bits, it's different with TDM choices I think.
Could someone explain in simple words the advantage of the different choices ?
Thanks,
gd_on

[brianpt]

Let me try the new compiler.  It is unbelievable that such a little bug can kill my code.