merged to master, thanks!
-jon
On 06/15/2018 12:05 AM, Matthew R. Trower wrote:
Jon Trulson writes:
I can add this, but for safety/clarity, can you change the comparison from:
#if __SUNPRO_CC == 0x5100
to
#if defined(__SUNPRO_CC) && __SUNPRO_CC == 0x5100
This is a little clearer and won't
Jon Trulson writes:
> I can add this, but for safety/clarity, can you change the comparison from:
>
> #if __SUNPRO_CC == 0x5100
>
> to
>
> #if defined(__SUNPRO_CC) && __SUNPRO_CC == 0x5100
>
> This is a little clearer and won't emit warnings under -Wundef on
> non-studio c compilers...
Certainly
On 06/13/2018 09:45 PM, Matthew R. Trower wrote:
d...@blackshard.net (Matthew R. Trower) writes:
I'll try to set up an environment with 12.3 tomorrow, and see what
happens.
Alright, so...
SolarisStudio12.3 compiles this fine. SunStudio12.1 encounters the
problem. It seems to me this is p
d...@blackshard.net (Matthew R. Trower) writes:
> I'll try to set up an environment with 12.3 tomorrow, and see what
> happens.
>
>
Alright, so...
SolarisStudio12.3 compiles this fine. SunStudio12.1 encounters the
problem. It seems to me this is probably a compiler bug, then.
However...
Poli
Ulrich Wilkens writes:
> On 06/11/18 04:30, Matthew R. Trower wrote:
>
>>
>> Nah. The issue is traceable back to Ptr.C and Ptr.h. Stripping out
>> preprocessor #if noise, the generated entmgr_inst.C tries to do this:
>>
>> ---
>>template class Ptr;
>>template class ConstPtr;
>> ---
>>
>
On 06/11/18 04:30, Matthew R. Trower wrote:
Nah. The issue is traceable back to Ptr.C and Ptr.h. Stripping out
preprocessor #if noise, the generated entmgr_inst.C tries to do this:
---
template class Ptr;
template class ConstPtr;
---
class Ptr defines and implements a 'clear()' method
d...@blackshard.net (Matthew R. Trower) writes:
> Here's a potential patch which solves this issue for me. It should, in
> my estimation, be functionally equivalent. Though, I'm still curious as
> to whether Ulrich gets a clean build without it, and if so, why.
>
> Thoughts?
Heck, even better..
Here's a potential patch which solves this issue for me. It should, in
my estimation, be functionally equivalent. Though, I'm still curious as
to whether Ulrich gets a clean build without it, and if so, why.
Thoughts?
-- Matthew R. Trower
>From 35f1b6ad5d76effa03793d3e8ff9bbd756c22e46 Mon S
Jon Trulson writes:
>
> That looks like the compiler can't tell the difference between a const
> and non-const definition. Perhaps you need to define:
>
> -DSP_VOLATILE=volatile -DSP_CONST=const
>
> for your architecture in programs/nsgmls/Imakefile like Linux and BSD.
>
> You might look an see
On 06/09/2018 05:23 PM, Matthew R. Trower wrote:
Build log attached.
Also attached is a WorksForMe patch, which I've been using to work
around the issue for some time. Apparently these extra definitions are
not a problem on other platforms, and presumedly there's some reason for
them to be th
10 matches
Mail list logo
