Adopting utl::log()

[Lucas-C]

No description provided.

[felixguendling]

I think we need an option in utl to enable the docs.
UTL_BUILD_DOCS with default value OFF.

And then wrap this in if (UTL_BUILD_DOCS):
https://github.com/motis-project/utl/blob/69e700e03ae3d385304d754cf57877528bc591ef/CMakeLists.txt#L33

[felixguendling]

Seeing the code, maybe it makes sense to rename to something with log in the name (e.g. utl::log_err)? At least utl::error could also be an exception class derived from std::exception.

[Lucas-C]

I think we need an option in utl to enable the docs. UTL_BUILD_DOCS with default value OFF.

And then wrap this in if (UTL_BUILD_DOCS): motis-project/utl@69e700e/CMakeLists.txt#L33

Do all the MOTIS repos builds rely on the CMakeLists.txt in the utl module?

I'm not a big CMake expert, so I'm not sure how best to share the doc-building logic between MOTIS repositories.
I think the following could be shared from motis-project/utl:

• docs/CMakeLists.txt
• docs/Doxyfile (except from the module name)
• docs/conf.py (except from the module name)
• docs/motis-logo.svg
• docs/requirements.txt
• the doc job from .github/workflows/unix.yml

Maybe this should be tackled in a separate issue / PR?

[Lucas-C]

Seeing the code, maybe it makes sense to rename to something with log in the name (e.g. utl::log_err)? At least utl::error could also be an exception class derived from std::exception.

Yes, I agree, I think it would be clearer.

I opened PR motis-project/utl#32 to add this prefix.

[felixguendling]

I think we need an option in utl to enable the docs. UTL_BUILD_DOCS with default value OFF.
And then wrap this in if (UTL_BUILD_DOCS): motis-project/utl@69e700e/CMakeLists.txt#L33

Do all the MOTIS repos builds rely on the CMakeLists.txt in the utl module?

I'm not a big CMake expert, so I'm not sure how best to share the doc-building logic between MOTIS repositories. I think the following could be shared from motis-project/utl:

• docs/CMakeLists.txt
• docs/Doxyfile (except from the module name)
• docs/conf.py (except from the module name)
• docs/motis-logo.svg
• docs/requirements.txt
• the doc job from .github/workflows/unix.yml

Maybe this should be tackled in a separate issue / PR?

In case this docs building is always the same and should be used in all MOTIS repositories (osr, motis, etc.), I would propose to create a separate repository (maybe called docs-gen or simply docs) and include it via .pkg. As long as it has a central CMakeLists.txt (then in the root directory) it will work. I think this is maybe a better solution than to force each repo to include utl if it wants only docs. Maybe at the moment all repos include utl but I would say it's maybe better to split it?

[Lucas-C]

I would propose to create a separate repository (maybe called docs-gen or simply docs) and include it via .pkg.
I think this is maybe a better solution than to force each repo to include utl if it wants only docs.
Maybe at the moment all repos include utl but I would say it's maybe better to split it?

I agree.
I will have to stop working on MOTIS for now, but I'll try to do this next week.
The documentation generation also requires Python and some depdencies to be installed with pip.
I'm not sure how best to do this from CMake, so unless you have some guidance regarding this, I'll search for existing projects doing something similar.

[Lucas-C]

Comment moved to: motis-project/motis#702

[felixguendling]

The purpose of tracing in RAPTOR include/nigiri/routing/raptor/debug.h is mainly to be dumped to a text file that can become gigabytes in size for a single earliest arrival query (not even range query!). We neither want this stuff to go OTEL nor do we need timestamps, log levels, etc. here. I would like to keep it fmt::println for this reason. It's for MOTIS developers who need to go very very deep and check millions of routing operations to find the needle in the haystack if results don't match what we expect (e.g. from previous versions, or other implementations).

[Lucas-C]

I am currently blocked by this issue: motis-project/utl#35

[felixguendling]

should be fixed by motis-project/utl#36

I am currently blocked by this issue: motis-project/utl#35

[Lucas-C]

I think this is ready

[felixguendling]

Why nigiri::scoped_timer if there's utl::scoped_timer?

[felixguendling]

I think you were right with the function signature. It's cluttering output a lot. Even if it gives additional context, it's maybe not worth to clutter the text output so much.