http://cheezburger.com/6361837056

I stumbled across two of this "bugs" in less than one hour.

Post code, that sounds interesting. Like, how does it even know what a string is? That information should have decayed. So it's either a compiler bug or a static assert gone awry*.






*or a false dichotomy

Am 23.06.2012 13:43, schrieb Bernard Helyer:
> Post code, that sounds interesting. Like, how does it even know what a
> string is? That information should have decayed. So it's either a
> compiler bug or a static assert gone awry*.
>
>
>
>
>
>
> *or a false dichotomy


The cool thing is, I wasn't able to track it down until now, since line numbers are completly messed up because of a heavy use of mixin() and CTFE.

> The cool thing is, I wasn't able to track it down until now, since line
> numbers are completly messed up because of a heavy use of mixin() and CTFE.


s/until now/so far/

Am 23.06.2012 13:49, schrieb David:
>> The cool thing is, I wasn't able to track it down until now, since line
>> numbers are completly messed up because of a heavy use of mixin() and
>> CTFE.
>
>
> s/until now/so far/

Seems like I did already fix that, https://github.com/D-Programming-Language/phobos/pull/482

Now I get linker errors, yay \o/

On Saturday, 23 June 2012 at 11:22:26 UTC, David wrote:
> http://cheezburger.com/6361837056
>
> I stumbled across two of this "bugs" in less than one hour.

I am currently writing a sat solver for educational purposes and hit a handful of bugs both in dmd and phobos. Still not able to reduce all of them, so that I can report them.

It sucks if you don't know whether it's a bug your you are doing something wrong with the language. At least they get fixed.

On Sat, Jun 23, 2012 at 10:58 PM, Tobias Pankrath <tobias@pankrath.net>wrote:

> I am currently writing a sat solver for educational purposes


How mature is the sat solver yet? Do you plan to release it at some point of time?

Regards
- Puneet

On Saturday, 23 June 2012 at 17:47:04 UTC, d coder wrote:
> On Sat, Jun 23, 2012 at 10:58 PM, Tobias Pankrath <tobias@pankrath.net>wrote:
>
>> I am currently writing a sat solver for educational purposes
>
>
> How mature is the sat solver yet? Do you plan to release it at some point
> of time?
>
> Regards
> - Puneet

It was a DPLL with simple BCP yesterday and is a non functional blob today :-).

I don't think I will push this to the state of the art. I plan for a conflict driver clause learning solver, that chooses variables like chaff and does non-chronological backtracking plus 2-watched-literal BCP.

IF I have enough time to do this. I'm new to sat and need to understand all of the above first :-).

If you need a mature one, it seems that minisat can be easily made available or ported to D. I'd do this, if it weren't for educational purposes.

Am 23.06.2012 19:28, schrieb Tobias Pankrath:
> On Saturday, 23 June 2012 at 11:22:26 UTC, David wrote:
>> http://cheezburger.com/6361837056
>>
>> I stumbled across two of this "bugs" in less than one hour.
>
> I am currently writing a sat solver for educational purposes and hit a
> handful of bugs both in dmd and phobos. Still not able to reduce all of
> them, so that I can report them.
>
> It sucks if you don't know whether it's a bug your you are doing
> something wrong with the language. At least they get fixed.


Yeah sometimes it's quite frustrating, but on the other hands you have those moments, fuck yeah it works! with a code which is unique to D, e.g. I was able to reduce a nearly 2k lines monster of a protocol implementation down to 300 lines with a heavy use of CTFE magic.

On Saturday, June 23, 2012 13:46:51 David wrote:
> The cool thing is, I wasn't able to track it down until now, since line numbers are completly messed up because of a heavy use of mixin() and CTFE.

Which is why I _never_ put newlines in string mixins.

- Jonathan M Davis