[C++][R] abseil fails to build with gcc-13

[thisisnic]

Describe the bug, including details regarding any error messages, version, and platform.

e.g. https://dev.azure.com/ursacomputing/crossbow/_build/results?buildId=52112&view=logs&j=0da5d1d9-276d-5173-c4c4-9d4d4ed14fdb&t=d9b15392-e4ce-5e4c-0c8c-b69645229181&l=1975

/tmp/RtmpL1MTjz/file4a454972d15/absl_ep-prefix/src/absl_ep/absl/strings/internal/str_format/extension.cc:45:1: note: in expansion of macro ‘ABSL_INTERNAL_CONVERSION_CHARS_EXPAND_’
   45 | ABSL_INTERNAL_CONVERSION_CHARS_EXPAND_(ABSL_INTERNAL_CHAR_SET_CASE, )
      | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/RtmpL1MTjz/file4a454972d15/absl_ep-prefix/src/absl_ep/absl/strings/internal/str_format/extension.cc:44:13: error: ‘FormatConversionCharSet’ does not name a type; did you mean ‘FormatConversionCharInternal’?
   44 |   constexpr FormatConversionCharSet FormatConversionCharSetInternal::c;
      |             ^~~~~~~~~~~~~~~~~~~~~~~

Component(s)

R

[thisisnic]

@assignUser - I don't suppose you've got any idea why this might have started failing?

[assignUser]

hm that reads like a version mismatch or something... I'll check it out

[assignUser]

I don't think this is abseil related as the files in question haven't had changes in a few years ...it looks like there was a major docker upgrade on azure from 20.x to 23.x

[assignUser]

looks like the r-base image bumped gcc from 12 to 13 that might be it?

[assignUser]

can reproduce the error locally with gcc 13, will investigate but as cran has multiple machines running 13 this is likely a blocker for the next 13 rc

HEAD abseil build without a problem so we might just need to bump the abseil version (which iirc is potentially difficult?)
cc @raulcd

[assignUser]

I think these are the relevant issues that are cause by gcc 13 https://gcc.gnu.org/gcc-13/porting_to.html#header-dep-changes

abseil-cpp/absl/strings/internal/str_format/extension.h:34:6: warning: elaborated-type-specifier for a scoped enum must not use the ‘class’ keyword
   34 | enum class FormatConversionChar : uint8_t;
      | ~~~~ ^~~~~
      |      -----
abseil-cpp/absl/strings/internal/str_format/extension.h:34:33: error: found ‘:’ in nested-name-specifier, expected ‘::’
   34 | enum class FormatConversionChar : uint8_t;
      |                                 ^
      |                                 ::

abseil-cpp/absl/strings/internal/str_format/extension.h:29:1: note: ‘uint8_t’ is defined in header ‘<cstdint>’; did you forget to ‘#include <cstdint>’?
   28 | #include "absl/strings/internal/str_format/output.h"
  +++ |+#include <cstdint>
   29 | #include "absl/strings/string_view.h"
abseil-cpp/absl/strings/internal/str_format/extension.h:187:29: error: default member initializer for unnamed bit-field
  187 |   enum class Enum : uint8_t {

[assignUser]

We actually don't have any CI build with gcc 13 (other than test-r-library-r-base-latest and test-r-depsource-bundled) so this is an issue not only affecting R but was missed thus far.

[h-vetinari]

@assignUser, is there a replacement issue, now that #37066 was closed unmerged? #37147 is a work-around, not a solution. Abseil should be kept up-to-date like any other dependency.

[assignUser]

With our move to cmake 3.16 as minumum version we can finally use FetchContent instead of ep, this will make updating dependencies much easier as we do not have to recreate the internal target tree of the dependency as we currently do.

It should also be possible to move singular dependencies to FC (caveat is that we have to move all of it's dependencies to FC as well afaik). I don't think there is a general issue for this or for abseil specifically.

[adamreeve]

This issue was blocking me from doing some development work so I changed the Abseil build to include a patch to work around it: adamreeve@ed44152

I don't think it would make sense to make a PR with this as the proper fix is to upgrade Abseil, but I thought I'd post it here in case it helps anyone else.