On Sunday, 5 August 2018 at 01:15:07 UTC, Nicholas Wilson wrote:
> On Saturday, 4 August 2018 at 18:12:05 UTC, Alex wrote:
>> On Saturday, 4 August 2018 at 13:26:01 UTC, Nicholas Wilson
> That is a very long stacks trace and combined with the very short stack trace on OSX, this is probably a stack overflow.
> 1) Do you have any very deeply nested types/templates somewhere or somehow created a cyclic dependency in semantic analysis?
> 2) Can you try with a bigger stack (ulimit -s)? If it still fails then its probably 1.

:)
I assumed this at the beginning too...
The structures are not very deep, say 3-4 levels. But yes, I do have something like a loop you mentioned.
I'm trying to get rid of this right now. I hoped, if the compiler has problems with my construct, it would help me to find a better start place for this...

However, if I take your description: Is there a general way, to solve a cyclic dependency during the semantic analysis step? I mean, I'm trying to lowering logic in my code somehow, but the loop I need won't go away per se...

On Sunday, 5 August 2018 at 10:12:39 UTC, Alex wrote:
> On Sunday, 5 August 2018 at 01:15:07 UTC, Nicholas Wilson wrote:
>> On Saturday, 4 August 2018 at 18:12:05 UTC, Alex wrote:
>>> On Saturday, 4 August 2018 at 13:26:01 UTC, Nicholas Wilson
>> That is a very long stacks trace and combined with the very short stack trace on OSX, this is probably a stack overflow.
>> 1) Do you have any very deeply nested types/templates somewhere or somehow created a cyclic dependency in semantic analysis?
>> 2) Can you try with a bigger stack (ulimit -s)? If it still fails then its probably 1.
>
> :)
> I assumed this at the beginning too...
> The structures are not very deep, say 3-4 levels. But yes, I do have something like a loop you mentioned.
> I'm trying to get rid of this right now. I hoped, if the compiler has problems with my construct, it would help me to find a better start place for this...
>
> However, if I take your description: Is there a general way, to solve a cyclic dependency during the semantic analysis step? I mean, I'm trying to lowering logic in my code somehow, but the loop I need won't go away per se...

I can't really know without seeing the code, but if you can move the body of the loop into its own template that might work.