Hi All
I am forced into retracting what I have said a few days ago:
"I have already email Walter about this issue and as he is slow as usual in addressing Linux back-end issues [...]"

Mea culpa.

Walter was very responsive (just to spite me? :)) and after a few nights of email exchanges and hacking that went well after 2AM, we found that:

1) the DW_AT_comp_dir piece of DWARF debug info, which gives the build path of a translation unit was being output correctly by the DMD compiler;

2) however, because it was in the DW_FORM_strp (offset to a string) form, rather than a plain DW_FORM_string, the linkers on some systems took the liberty to mess with it (see http://sourceware.org/ml/gdb/2002-04/msg00144.html:

"Some linkers (including the GNU linker) can remove duplicate strings from an executable's .debug_str section, and adjust DW_FORM_strp references to point to the remaining unique copies.")

3) The resolution was to use DW_FORM_string, and everything works correctly now.


Also, we have agreed upon the debug formats for dynamic D arrays and associative arrays.

I have implemented support for the former in ZeroBUGS, and it will go out soon.

After I get the new builds out, I will start looking into debugger support for associative arrays (although I cannot promise to deliver anything in the very near future, due to expecting a new family addition any minute now).

cheers,
   Cristian Vlasceanu (http://www.zerobugs.org)

Cristian Vlasceanu wrote:
> Hi All
> I am forced into retracting what I have said a few days ago:
> "I have already email Walter about this issue and as he is slow as usual in addressing Linux back-end issues [...]"
> 
> Mea culpa.
> 

That (the above) is one of the reasons I like D.

> Walter was very responsive (just to spite me? :)) and after a few nights of email exchanges and hacking that went well after 2AM, we found that:
> 
> 1) the DW_AT_comp_dir piece of DWARF debug info, which gives the build path of a translation unit was being output correctly by the DMD compiler;
> 
> 2) however, because it was in the DW_FORM_strp (offset to a string) form, rather than a plain DW_FORM_string, the linkers on some systems took the liberty to mess with it (see http://sourceware.org/ml/gdb/2002-04/msg00144.html:
> 
> "Some linkers (including the GNU linker) can remove duplicate strings from an executable's .debug_str section, and adjust DW_FORM_strp references to point to the remaining unique copies.")
> 
> 3) The resolution was to use DW_FORM_string, and everything works correctly now.
> 

Sweet

> 
> Also, we have agreed upon the debug formats for dynamic D arrays and associative arrays.
> 

Again

> I have implemented support for the former in ZeroBUGS, and it will go out soon.
>

I assume that DMD 1.011 will generate said new debug stuff??

> After I get the new builds out, I will start looking into debugger support for associative arrays (although I cannot promise to deliver anything in the very near future, due to expecting a new family addition any minute now).
> 

congrats x2!

> cheers,
>    Cristian Vlasceanu (http://www.zerobugs.org)

Cristian Vlasceanu wrote:
> 
> After I get the new builds out, I will start looking into debugger support for associative arrays (although I cannot promise to deliver anything in the very near future, due to expecting a new family addition any minute now).
> 
> cheers,
>    Cristian Vlasceanu (http://www.zerobugs.org)

sounds nice but .. no windows? :'(

Hasan Aljudy wrote:

> 
> 
> Cristian Vlasceanu wrote:
>> 
>> After I get the new builds out, I will start looking into debugger support for associative arrays (although I cannot promise to deliver anything in the very near future, due to expecting a new family addition any minute now).
>> 
>> cheers,
>>    Cristian Vlasceanu (http://www.zerobugs.org)
> 
> sounds nice but .. no windows? :'(

AFAIK ddbg is for Windows?

-- 
Lars Ivar Igesund
blog at http://larsivi.net
DSource, #d.tango & #D: larsivi
Dancing the Tango

the upcoming Ddbg release takes advantage of the DMD 1.011 debugging features.

Lars Ivar Igesund wrote:
> Hasan Aljudy wrote:
> 
>>
>> Cristian Vlasceanu wrote:
>>> After I get the new builds out, I will start looking into debugger support for associative arrays (although I cannot promise to deliver anything in the very near future, due to expecting a new family addition any minute now).
>>>
>>> cheers,
>>>    Cristian Vlasceanu (http://www.zerobugs.org)
>> sounds nice but .. no windows? :'(
> 
> AFAIK ddbg is for Windows?
> 

Lars Ivar Igesund wrote:
> Hasan Aljudy wrote:
> 
>>
>> Cristian Vlasceanu wrote:
>>> After I get the new builds out, I will start looking into debugger
>>> support for associative arrays (although I cannot promise to deliver
>>> anything in the very near future, due to expecting a new family addition
>>> any minute now).
>>>
>>> cheers,
>>>    Cristian Vlasceanu (http://www.zerobugs.org)
>> sounds nice but .. no windows? :'(
> 
> AFAIK ddbg is for Windows?
> 

I know, I already use it! but the more the better! :P

Reply to Jascha,

> the upcoming Ddbg release takes advantage of the DMD 1.011 debugging
> features.

wow, people around here ate totally amazing they get support for new features working before the new features are working!

Did Walter push out some private alpha versions of 1.011?

BCS wrote:
> Reply to Jascha,
> 
>> the upcoming Ddbg release takes advantage of the DMD 1.011 debugging
>> features.
> 
> wow, people around here ate totally amazing they get support for new features working before the new features are working!
> 
> Did Walter push out some private alpha versions of 1.011?

I don't think he'd have to. If the format of the debugging info DMD will output is specified, the debugger can be made to parse it without needing to have access to a compiler that actually outputs it...