> We have to catch an error of byte-compilation,
To what end? The individual can see build errors and I presume Hydra
reports build failures originating in the Makefile.
> but there already exists this target and rule, and I'd like not to
> change the original behavior.
Doing in ert.el what the Makefile is responsible for seems off to me.
If test code build errors are to go in the summary, why shouldn't
build errors of the code under test? If incorporating build failures
into a summary report is good enough to do, why isn't it good enough
for 'make check'?