[PATCH] D113717: [Symbolizer][Debuginfo] Add debuginfod client to llvm-symbolizer.

David Blaikie via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Dec 13 16:57:08 PST 2021


dblaikie added a comment.

In D113717#3189211 <https://reviews.llvm.org/D113717#3189211>, @thakis wrote:

>> I'm not 100% sure that it's obvious the linker shouldn't make http requests if it can be used to improve diagnostics.
>
> It's very scary to me that you say that. Maybe we should have a in-person (well, VC) discussion about this change?

Happy to, if you like - my calendar's usually pretty open - either scheduled or impromptu is fine by me.

> To me, separation of concerns makes it fairly obvious that compilers and linkers etc shouldn't rely on an internet connection to work. There should be a separate binary that handles downloading stuff that can be run well in advance (or, not run). Kitchen-sync binaries/libraries etc tend to be not great to work with.
>
> D115554 <https://reviews.llvm.org/D115554> looks fine to me, independent of this change. I don't think it helps this change here much since another tool might add an accidental (or even intentional) dependency on DebugInfo/Symbollize again, like lld did in the past.
>
> D115500 <https://reviews.llvm.org/D115500> also looks fine to me in isolation, but it doesn't help with the overall design here.
>
> I fairly strongy feel that there should be a library for just reading this debuginfod cache that doesn't do web requests that all tools could link to without concerns, and a separate library for writing to it and doing network requests, and that should be linked to from very few binaries.




CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D113717/new/

https://reviews.llvm.org/D113717



More information about the llvm-commits mailing list