Now I have some more positive responses below: On 11/9/07, Marc Lehmann <[EMAIL PROTECTED]> wrote: > On Fri, Nov 09, 2007 at 01:19:02PM +0100, Chris Brody-GMail <[EMAIL > PROTECTED]> wrote: > > with the following flags to workaround a priority event issue: > > -Dev_flags=1 -DEVLIST_ACTIVE=1. This both solves a compile error and > > (omg, this works????!?!?!?!) :) > > Now, I always considered mucking with the flags as bad style (its internal > and should not be used by apps), but of course its in the header file and > not being documented means nothing. My only reason to define these flags was: these are not defined in libev yet and had broken the build of eventxx. If I had not defined these as "=1", (g)cc should do this by default. It just happened that these particular definition values would cause the function to abort, which is correct for missing functionality.
I wanted to get this fixed with as few local changes as possible. Despite the badness of putting such flags in a makefile, this allows me to keep going with a minimum of personal code merges until we are ready with the proper priority support. > > First I got the following errors: > > ./eventxx: In destructor `eventxx::dispatcher::~dispatcher()': > > ./eventxx:794: error: cannot convert `eventxx::internal::event_base*' > > to `event_base*' for argument `1' to `void > > eventxx::internal::event_base_free(event_base*)' > > yeah, thats a known bug, but I wondered how many compilers would get it wrong > when writing that code. Right-I may try this again on my OSX gcc 4.0, but I'd rather keep going with working code than start debugging compilers right now... > > Very strange, but by declaring struct event_base at the beginning of > > Thats indeed very weird. No, I have no problems whatsoever with that > change (its actually much cleaner that way, I think), thanks for tracking > it down, it'll be in CVS in a few hours.' Perfect-thanks again for the support-always make me happy to see alternatives working. I only took a quick look at your remarks on C++ so far, but I like both your analysis of the C++ objects and of the legal junk. I tend to prefer to remain on the safer side-for example, I follow the Mundy/Microsoft idea to avoid reading (L)GPL source code, to avoid the possiblity of accidentally copy by reuse of restricted source code. But I agree to let these things be sorted out by the real legal experts... One more quick remark on C++: I also like to avoid templated code, primarily due to the frequency of compiler bugs. As you know, I don't always use the latest GCC for example, so again I kindof like the safe side here. CB _______________________________________________ Libevent-users mailing list Libevent-users@monkey.org http://monkeymail.org/mailman/listinfo/libevent-users
