Building ATS with clang

[Igor Galić <i....@brainsware.org>]

Hi folks,

in discussing clang with zwoop on IRC, I ended up doing a port
this weekend. The progress is documented here:

https://issues.apache.org/jira/browse/TS-427

The good news is that it now compiles with Clang. The port
can also be compiled with GCC - and when compiled with GCC
it also ``passes'' the regression test (i.e.: I'm not 
introducing any more regressions than there already are ;)

The bad news is that it doesn't pass the regression test
with clang: It will hang in ``Regression test(SDK_API_INKCache''
(see attachment of the strace).

I believe that it does so for the same reason it abort()s the
regression test, when compiled clang and --enable-debug.
So far I haven't been able to track that down. (You can find
the stack trace in Jira too.)

If this patchset is to be accepted, or even considered, it needs
review and testing with different compilers. I'm especially
suspect of (old versions of) VC++, thanks to certain comments
in the code.

In essence this is a clean-up, or the first step to it, anyway:
The *print() calls need a review, or rather: A wrapper like APR
has, as to not mess with format strings.

The stuff that's really disturbing is proxy/mgt2 with cli and cli2,
as well as proxy/logging/LogBufferV1.h -- but this might be due
to my lack of insight in the code.

Thoughts?

i

-- 
Igor Galić

Tel: +43 (0) 664 886 22 883
Mail: i.galic@brainsware.org
URL: http://brainsware.org/

Re: Building ATS with clang

[Leif Hedstrom <zw...@apache.org>]

  On 08/28/2010 04:34 PM, Igor Galić wrote:
> Hi folks,
>
> in discussing clang with zwoop on IRC, I ended up doing a port
> this weekend. The progress is documented here:
>
> https://issues.apache.org/jira/browse/TS-427

Very cool. I won't have time to look at this for the next week or so, 
due to "work issues", hopefully someone else can take a look at it ?


>
> If this patchset is to be accepted, or even considered, it needs
> review and testing with different compilers. I'm especially
> suspect of (old versions of) VC++, thanks to certain comments
> in the code.

That is a none issue I think, since we don't support Windows. The VC++ 
comments are legacy, and I wouldn't be too concerned about them 
(although, intentionally breaking it for no good reason would seem 
counter productive :).  If (or when) someone decides to update the code 
to run on Windows, we can revisit), so maybe making a // TODO  comment 
could also be useful for future developers.


Cheers!

-- leif