Could you tell me (point me to source code) how does the Swift-CI
integration on GitHub piece together the benchmark report? I'd like to fix
the subheadings "Unoptimized (Onone)" and "Hardware Overview" that are not
being displayed correctly. I suggest we use ### Subheading instead of
trying to **bold** the text.
I'm also toying with a fix for overflowing results table by inserting
breaks in our ever longer test names in the comp_perf_test.py. The
situation is particularly egregious in GitHub's mobile view:
IMO this would considerably improve the situation. The only downside is
that if you copied the test names, you would not be able to find them in
the project, because of the invisible characters inserted between the came
case word boundaries. What do you think?
Could you tell me (point me to source code) how does the Swift-CI integration on GitHub piece together the benchmark report? I'd like to fix the subheadings "Unoptimized (Onone)" and "Hardware Overview" that are not being displayed correctly. I suggest we use ### Subheading instead of trying to **bold** the text.
This is part of Jenkins CI config, I will update the config to use ###.
Hello Mishal, hello swift-dev
Could you tell me (point me to source code) how does the Swift-CI integration on GitHub piece together the benchmark report? I'd like to fix the subheadings "Unoptimized (Onone)" and "Hardware Overview" that are not being displayed correctly. I suggest we use ### Subheading instead of trying to **bold** the text.
I'm also toying with a fix for overflowing results table by inserting breaks in our ever longer test names in the comp_perf_test.py. The situation is particularly egregious in GitHub's mobile view:
<File 25-05-2017, 23 07 35.png>
I could use soft hypens or zero-width spaces:
Soft Hyphenated
TEST MIN MAX MEAN MAX_RSS
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer1 664663 664928 664810 9409877
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer2 943681 943979 943869 9418069
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer3 1252577 1254327 1253654 9412608
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer4 933119 933913 933470 9415339
Zero-width Spaced
TEST MIN MAX MEAN MAX_RSS
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer1 664663 664928 664810 9409877
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer2 943681 943979 943869 9418069
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer3 1252577 1254327 1253654 9412608
ExistentialTestPassExistentialOneMethodCall_ClassValueBuffer4 933119 933913 933470 9415339
IMO this would considerably improve the situation. The only downside is that if you copied the test names, you would not be able to find them in the project, because of the invisible characters inserted between the came case word boundaries. What do you think?
I think zero-with spaced might be better, because if test contains “-“ it might cause confusion.
···
On May 25, 2017, at 2:14 PM, Pavol Vaskovic <pali@pali.sk> wrote: