[roottest] examine the status of h2root on Windows

[ferdymercury]

This Pull request:

Changes or fixes:

Check if h2root works on Windows.

if yes, @bellenot can you run the h2root test in a Debug build of Windows?

[github-actions]

Test Results

    20 files      20 suites   3d 14h 17m 37s ⏱️
 3 045 tests  3 045 ✅ 0 💤 0 ❌
59 310 runs  59 310 ✅ 0 💤 0 ❌

Results for commit da78596.

[ferdymercury]

I see on Windows CI:
2025-06-26T05:50:00.1510211Z -- Looking for a Fortran compiler - NOTFOUND

could you check locally? or install gfortran on that Windows image?
Thanks!

[guitargeek]

Classic. That's why we like fail-on-missing:

• [CMake] Also consider fail-on-missing for fortran features #19188

[bellenot]

I see on Windows CI: 2025-06-26T05:50:00.1510211Z -- Looking for a Fortran compiler - NOTFOUND

could you check locally? or install gfortran on that Windows image? Thanks!

FYI, there is no gfortran for Windows

[ferdymercury]

I see on Windows CI: 2025-06-26T05:50:00.1510211Z -- Looking for a Fortran compiler - NOTFOUND
could you check locally? or install gfortran on that Windows image? Thanks!

FYI, there is no gfortran for Windows

Right, sorry, the Intel-Fortran compiler.

[bellenot]

I see on Windows CI: 2025-06-26T05:50:00.1510211Z -- Looking for a Fortran compiler - NOTFOUND
could you check locally? or install gfortran on that Windows image? Thanks!

FYI, there is no gfortran for Windows

Right, sorry, the Intel-Fortran compiler.

It doesn't work. I guess I should try to fix it, but what is the priority?

[ferdymercury]

It doesn't work. I guess I should try to fix it, but what is the priority?

Low priority. One question though: is it h2root failing, or is the it just not compiling / the fortran compiler giving errors?

My hunch was that sometimes the Intel compilers discover bugs, so maybe the errors that Windows give can help in debugging the failures on MacOs.

[guitargeek]

Yes it's not important to get it to work on WIndows per se, but these errors on macOS are quite worrying. I think h2root needs to be reliable, otherwise we risk to corrupt data from the LEP era when converting to ROOT :(

[bellenot]

It doesn't work. I guess I should try to fix it, but what is the priority?

Low priority. One question though: is it h2root failing, or is the it just not compiling / the fortran compiler giving errors?

Short: CMake didn't find the compiler, and when specifying it, the compileer didn't pass the CMake tests at configure time. But I'll give it another try today

[bellenot]

So FYI I tried hard the whole day to build with Intel Fortran compiler, but it mess up the environment for C++, and mixing languages don't work, with this kind of error:

Compiling with Intel® Fortran Compiler 2025.2.0 [Intel(R) 64]...
g2root.f
ifx: command line warning #10161: unrecognized source type 'Files\Microsoft'; object file assumed
ifx: command line warning #10161: unrecognized source type 'Visual'; object file assumed
fpp: fatal: can't fopen file: Studio\2022\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64 /Qm64 C:\root-dev\git\master\main\src\g2root.f

I'll try to debug this next week, but I'm not very optimistic...

[dpiparo]

we can also decide not to provide h2root on windows.

[dpiparo]

we can also decide not to provide h2root on windows. If one wishes to play with lep data and convert them to ROOT format on Windows, wsl is available.

[ferdymercury]

Yep, I agree. This was more an attempt to see if compiling h2root with Intel gave more insightful warnings that helped to debug the macOS crashes. (Things such as access after end of list or so).

[ferdymercury]

So FYI I tried hard the whole day to build with Intel Fortran compiler, but it mess up the environment for C++, and mixing languages don't work, with this kind of error:

Compiling with Intel® Fortran Compiler 2025.2.0 [Intel(R) 64]...
g2root.f
ifx: command line warning #10161: unrecognized source type 'Files\Microsoft'; object file assumed
ifx: command line warning #10161: unrecognized source type 'Visual'; object file assumed
fpp: fatal: can't fopen file: Studio\2022\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64 /Qm64 C:\root-dev\git\master\main\src\g2root.f

I'll try to debug this next week, but I'm not very optimistic...

Thanks! It seems ifx does not like whitespaces in the compiler folder name. Any way to rename that folder, or to escape those whitespaces?

[bellenot]

Thanks! It seems ifx does not like whitespaces in the compiler folder name. Any way to rename that folder, or to escape those whitespaces?

I'll try to figure out and find a workaround. The issue is the inclusion with Visual Studio (hence the PATH - and without that it doesn't work)

[bellenot]

we can also decide not to provide h2root on windows. If one wishes to play with lep data and convert them to ROOT format on Windows, wsl is available.

We don't provide h2root on Windows. It was working in the past with ROOT 5 when using makefiles in the cygwin environmemt, but never tried with CMake, and this is the major issue I have now...

[ferdymercury]

Thanks a lot for checking. In that case, would it work to compile it on Windows using the old Makefiles from ROOT5 ? (Or just via the command line by hand, compiling the involved files one after the other?) Since I was just interested in seeing if if it raises any compiler warnings when compiling zebra.f and hbook.f especially.

[bellenot]

No warning at all. But I didn't find how to specify the warning level... Looking into it now

[bellenot]

OK, I found /warn:all, but then it is very verbose...

[bellenot]

@ferdymercury no relevant warnings, only plenty of warnings like:

C:\root-dev\git\master\misc\minicern\src\zebra.f(7213): warning #8889: Explicit interface or EXTERNAL declaration is required.   [MZGARB]
      CALL MZGARB (0,IXWIPE)
-----------^
[...]
--------------------------------------------^
C:\root-dev\git\master\misc\minicern\src\zebra.f(7207): warning #6717: This name has not been given an explicit type.   [IQVREM]
      COMMON /ZVFAUT/IQVID(2),IQVSTA,IQVLOG,IQVTHR(2),IQVREM(2,6)
------------------------------------------------------^
C:\root-dev\git\master\misc\minicern\src\zebra.f(7209): remark #7712: This variable has not been used.   [NAMESR]
      DIMENSION    NAMESR(2)
-------------------^

[ferdymercury]

Thanks a lot! Did you remove the -w flag? (see d8215a8) Or maybe it does not apply to Windows.

If possible, please attach a zip file with the warnings :)

[bellenot]

Here are the logs
ifx-logs.zip

[ferdymercury]

Thanks! Closing for the moment.