This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH v2 4/4] Add documentation about gengtype and inheritance


On 10/23/13 21:00, David Malcolm wrote:

I went through the various requirements listed in the doc and
investigated what happens when they're violated.  Some lead to immediate
build failure in gengtype (which is relatively sane), but some lead to
silent lack of field-traversal.  The details:
Yea, build failure in gengtype is reasonable. Build failure because gengtype generated bogons, is slightly less desirable, but still in the realm of OK IMHO. Silent lack of field traversal is bad bad bad.



Summarizing:
(B), (F), (G): silent possibility of failing to visit fields during
traversal (bad, bad, bad)
These are obviously the most worrisome.



(A),  (C), (D), (E): build-time failures (relatively benign by
comparison)

As for this patch, it is OK once the prerequisites go in.

AIUI you've approved each patch in the whole series, and so far I've
committed patches 1 and 2 of the 4 (relatively safe).

Clearly I should add some robustness to better handle the more egregious
failure modes described above.  Should I post patches to do so before
committing further, or can I commit the rest of the approved patches in
the "v2" series (patches 3 and 4), doing the hardening patches as a
followup?
Your call as long as you commit to do what you can to harden this code.


As an aside, has anyone tried writing a testsuite for gengtype and the
GTY machinery?
Not that I'm aware of.

jeff


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]