Re: nogc.cc & darwin

[Bryce McKinlay <bryce at waitaki dot otago dot ac dot nz>]

Andreas Tobler wrote:

>Second, two functions are not defined in nogc.cc, they are used in java/lang/ref/natReference.cc.
>
>_Jv_GCCanReclaimSoftReference (jobject) and
>_Jv_GCRegisterDisappearingLink (jobject *objp)
>
>I implemented them as 'null' functions, copied from boehm.cc. Is this
>correct? 
>
Yes.

>This should not be a patch, more a RFC. I couldn't test now. If the way
>is ok I'll deliver a patch.
>
>Third, I have an undefined symbol where I struggle as well. When linking
>I miss the symbol _sqrtf. My search led to the point that this
>symbol/function is in libstdc++.a but this one isn't on the linking libs
>list. Well, I didn't say --disabled-shared because I assumed that this
>is default on darwin.
>Do I have explicitly say disable-shared?
>

I don't think so. sqrtf should be defined by java/lang/w_sqrt.o in 
libjava. libgcj doesn't need libstdc++ so you shouldn't need to link it.

>And last, how about patches, I see different formats diff -rcs or diff
>-u, which one is preferred? I prefer -u since it uses less space, but
>I'm open to do as expected.
>

Some people like -c (context diff) but -u (unified diff) seems to be the 
most popular for GCC. So, -u is fine.

>And finally the diff, as said, this one should not be applied since I
>couldn't test so far and I'd like to hear your opinion.
>

It looks fine. nogc.cc is only used occasionally when trying to get a 
new port up and running, which is why it is broken so often. Any patch 
to fix it is welcome ;-)

regards

Bryce.