Re: Is gcj dead?

On Tue, 20 Oct 2009, Joel Dice wrote:

> On Tue, 20 Oct 2009, Andrew Haley wrote:
>
>> Joel Dice wrote:
>>> On Tue, 20 Oct 2009, Andrew Haley wrote:
>>> 
>>>> Joel Dice wrote:
>>>>> On Mon, 19 Oct 2009, Andrew Haley wrote:
>>>>> 
>>>>>> Mathieu Malaterre wrote:
>>>>>>> On Mon, Oct 19, 2009 at 3:10 PM, Andrew Haley <[email protected]> wrote:
>>>>>>>> Mathieu Malaterre wrote:
>>>>>>>>> On Mon, Oct 19, 2009 at 3:00 PM, Andrew Haley <[email protected]>
>>>>>>>>> wrote:
>>>>>>>>>> Mathieu Malaterre wrote:
>>>>>>>>>>> On Sun, Oct 18, 2009 at 12:08 PM, Andrew Haley <[email protected]>
>>>>>>>>>>> wrote:
>>>>>>>>>>>> Yuri wrote:
>>>>>>>>>>>>> Last news in http://gcc.gnu.org/java/ are dated March 2007.
>>>>>>>>>>>> Yes, we should update that.  There hasn't been a lot of  new gcj
>>>>>>>>>>>> development,
>>>>>>>>>>>> but it is maintained.
>>>>>>>>>>>> 
>>>>>>>>>>>>> Also I submitted few PRs a month ago and there is no response
>>>>>>>>>>>>> at all.
>>>>>>>>>>>> Which ones?
>>>>>>>>>>> How about this one:
>>>>>>>>>>> 
>>>>>>>>>>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40816
>>>>>>>>>> I am still rather nervous about that one, as it's an ABI change.
>>>>>>>>> Point taken.
>>>>>>>>> In the long term this will prevent compilation of package such as
>>>>>>>>> VTK
>>>>>>>>> on debian on arch such as HPPA.
>>>>>>>> Really?  That's all rather amazing.  Is there no simple workaround?
>>>>>>> 
>>>>>>> Compilation error can be found here:
>>>>>>> 
>>>>>>> http://www.vtk.org/pipermail/vtk-developers/2009-June/006110.html
>>>>>>> 
>>>>>>> And source:
>>>>>>> 
>>>>>>> http://public.kitware.com/cgi-bin/viewcvs.cgi/Graphics/vtkJVMManager.h?view=annotate
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> I really do not see how I can work around that. Simply removing one of
>>>>>>> the multiple signature is not a solution IMHO.
>>>>>> 
>>>>>> Yes, I see what's going on.
>>>>>> 
>>>>>> To Tom Tromey: This is an ABI change, but AFAICS the only time it makes
>>>>>> a difference is where it's already broken.  I'm tempted to make the
>>>>>> change
>>>>>> now.
>>>>> 
>>>>> On the subject of ABI bugs, perhaps this patch is also ready for prime
>>>>> time:
>>>>> 
>>>>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28474
>>>> 
>>>> I don't understand the comment
>>>> 
>>>> "I will attach a patch which fixes the first case but not the second
>>>> (since I'm
>>>> not sure how the second case was intended to be handled)."
>>>> 
>>>> What second case is that?
>>> 
>>> Sorry it's not clear.  The "second case" refers to the xy__User symbol
>>> in the example program, which is not fixed by the patch because it
>>> includes the sequence "__U".  Only the first case, in which the
>>> underscores and "U" are not consecutive, is fixed.
>> 
>> OK.
>> 
>>>> BTW, if this patch had been submitted to java-patches at the time it
>>>> would have
>>>> gone straight in.
>>> 
>>> I submitted it to [email protected] (at your request) over two
>>> years ago:
>>> 
>>> http://gcc.gnu.org/ml/java/2007-04/msg00007.html
>>> http://gcc.gnu.org/ml/gcc-patches/2007-04/msg00015.html
>>> 
>>> I didn't realize that it also needed to go to java-patches.
>> 
>> Oh no, I'm really sorry.  This was my fault, I simply forgot.
>
> No problem - I forgot about it, too, until now :)
>
>> It seems that I was waiting for a complete patch, but one never arrived.
>> Any patch that goes in now really must be correct and complete, though.
>
> Agreed.  My (possibly naive) suggestion would be to use an escape character 
> or sequence which cannot appear in Java symbol ($$U, perhaps?).

Actually, "$$U" may not be ideal - the Java spec permits "$" in 
identifiers (but recommends only using it for mechanically generated code 
and legacy compatibility).

>> I'll have a look now.
>> 
>> Andrew.