clang-format (and related code changes)

[valassi]

Hi @roiser, this is my first draft proposal for clang-format (see #49).

Done:

• completed .clang-format
• formatted ggttgg code using that .clang-format (and additional manual changes including very few on/off blocks)

To do:

• test that the code builds and runs (not yet done...)
• backport to code generation
• regenerate all processes
• add to my generation scripts some checks that the clang-format is being respected
• (optionally: for someone else? add a merge hook to check that the code is respected?)

Have a look if you have comments! Thanks Andrea

[valassi]

I have now done

• fixed a functional bug (I had hardcoded a color=24)
• regenerated all 5 main processes, did a first basic test of performance
• analysed and fixed a performance regression in eemumu: I had added a few extra parentheses as "( TMP4 ) + ( - cI )" as a workaround to avoid expressions like "( TMP4 ) - cI" that clang-format (BUG?!) interprets as a C cast "(TMP4)-cI", and this and similar issues were adding a performance overhead; now fixed as "TMP4 - cI" in code generation, recovering performance

In progress

• other non SM processes

TODO

• complete non SM processes
• regenerate all 5 processes with final formatting, and rerun all performance tests

[valassi]

About other processes, a few issues to fix:

diff bb_tt.auto/SubProcesses/P1_Sigma_sm_bbx_ttx/CPPProcess.cc.badfmt bb_tt.auto/SubProcesses/P1_Sigma_sm_bbx_ttx/CPPProcess.cc
171,172c171,172
<       jamp_sv[0] += +1. / 6. * amp_sv[0];
<       jamp_sv[1] -= 1. / 2. * amp_sv[0];
---
>       jamp_sv[0] += +1./6. * amp_sv[0];
>       jamp_sv[1] -= 1./2. * amp_sv[0];

diff pp_tt.auto/SubProcesses/P1_Sigma_sm_uux_ttx/CPPProcess.cc pp_tt.auto/SubProcesses/P1_Sigma_sm_uux_ttx/CPPProcess.cc.badfmt 
181,182c181,182
<       jamp_sv[0] += +1./6. * amp_sv[0];
<       jamp_sv[1] -= 1./2. * amp_sv[0];
---
>       jamp_sv[0] += +1. / 6. * amp_sv[0];
>       jamp_sv[1] -= 1. / 2. * amp_sv[0];
511c511
<     const int denominators = 36,36; // FIXME: assume process.nprocesses == 1 for the moment (eventually denominators[nprocesses]?)
---
>     const int denominators = 36, 36; // FIXME: assume process.nprocesses == 1 for the moment (eventually denominators[nprocesses]?)

diff pp_tt.auto/SubProcesses/P1_Sigma_sm_uux_ttx/CPPProcess.h pp_tt.auto/SubProcesses/P1_Sigma_sm_uux_ttx/CPPProcess.h.badfmt 
30,32c30,32
< // Process: c c~ > t t~ WEIGHTED<=2 @1
< // Process: d d~ > t t~ WEIGHTED<=2 @1
< // Process: s s~ > t t~ WEIGHTED<=2 @1
---
>   // Process: c c~ > t t~ WEIGHTED<=2 @1
>   // Process: d d~ > t t~ WEIGHTED<=2 @1
>   // Process: s s~ > t t~ WEIGHTED<=2 @1

diff uu_dd.auto/SubProcesses/P1_Sigma_sm_uux_ddx/CPPProcess.cc uu_dd.auto/SubProcesses/P1_Sigma_sm_uux_ddx/CPPProcess.cc.badfmt 
173,174c173,174
<       jamp_sv[0] += +1./6. * amp_sv[0];
<       jamp_sv[1] -= 1./2. * amp_sv[0];
---
>       jamp_sv[0] += +1. / 6. * amp_sv[0];
>       jamp_sv[1] -= 1. / 2. * amp_sv[0];

diff uu_tt.auto/SubProcesses/P1_Sigma_sm_uux_ttx/CPPProcess.cc uu_tt.auto/SubProcesses/P1_Sigma_sm_uux_ttx/CPPProcess.cc.badfmt 
178,179c178,179
<       jamp_sv[0] += +1./6. * amp_sv[0];
<       jamp_sv[1] -= 1./2. * amp_sv[0];
---
>       jamp_sv[0] += +1. / 6. * amp_sv[0];
>       jamp_sv[1] -= 1. / 2. * amp_sv[0];

diff heft_gg_h.auto/SubProcesses/P1_Sigma_heft_gg_h/CPPProcess.cc heft_gg_h.auto/SubProcesses/P1_Sigma_heft_gg_h/CPPProcess.cc.badfmt 
160d159
< 

[valassi]

Done:

• I have fixed all pending issues in the "other" processes. The only pending issue is the "const int denominators = 36,36;" in pptt, but this is a known issue Fix "nprocesses>1" code generation (example: uutt within pptt gives "const int denominators = 36,36;") #343 which must be fixed by upgrading to a more recent upstream version Port epochX CODEGEN from 270gpu to the 311lovec branch (and pick up the new features there!) #360
• I have rerun a standard set of tests for the main five processes. All looks ok, performance is now the same or a bit better than it used to be (possibly because I managed to remove some parantheses and + signs, rather than adding a few more)

This is now ready. I will self merge this.