www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Anyone using nogc exceptions (-dip1008)?

reply Mike Parker <aldacron gmail.com> writes:
Some time ago, Walter asked me to postpone DIP 1008, "Exceptions 
and  nogc". Subsequently, an implementation was merged behind the 
-dip1008 flag. In helping to determine how to move forward with 
the DIP, we'd like to know if anyone is actually using the 
feature and, if so, what your experience with it has been.

And a small request, if you feel the urge to say something about 
the DIP process in this thread, please start a new thread 
instead. I'd like to keep this one on topic so I don't have to 
dig for the relevant info.

Thanks!
Mar 06 2019
next sibling parent reply Nicholas Wilson <iamthewilsonator hotmail.com> writes:
On Wednesday, 6 March 2019 at 12:36:56 UTC, Mike Parker wrote:

 Some time ago, Walter asked me to postpone DIP 1008, 
 "Exceptions and  nogc". Subsequently, an implementation was 
 merged behind the -dip1008 flag. In helping to determine how to 
 move forward with the DIP, we'd like to know if anyone is 
 actually using the feature and, if so, what your experience 
 with it has been.


No, because It doesn't work.

https://issues.dlang.org/show_bug.cgi?id=19317
https://issues.dlang.org/show_bug.cgi?id=19463
Mar 06 2019
parent Walter Bright <newshound2 digitalmars.com> writes:
On 3/6/2019 4:45 AM, Nicholas Wilson wrote:

 https://issues.dlang.org/show_bug.cgi?id=19463


https://github.com/dlang/dmd/pull/9481
Mar 23 2019
prev sibling next sibling parent Guillaume Piolat <first.last gmail.com> writes:
On Wednesday, 6 March 2019 at 12:36:56 UTC, Mike Parker wrote:

 Some time ago, Walter asked me to postpone DIP 1008, 
 "Exceptions and  nogc". Subsequently, an implementation was 
 merged behind the -dip1008 flag. In helping to determine how to 
 move forward with the DIP, we'd like to know if anyone is 
 actually using the feature and, if so, what your experience 
 with it has been.

 And a small request, if you feel the urge to say something 
 about the DIP process in this thread, please start a new thread 
 instead. I'd like to keep this one on topic so I don't have to 
 dig for the relevant info.

 Thanks!


Haven't used it, however we could have a use for it one day as we 
throw manually allocated  nogc exception.
Mar 06 2019
prev sibling next sibling parent lagfra <me fragal.eu> writes:
On Wednesday, 6 March 2019 at 12:36:56 UTC, Mike Parker wrote:

 Some time ago, Walter asked me to postpone DIP 1008, 
 "Exceptions and  nogc". Subsequently, an implementation was 
 merged behind the -dip1008 flag. In helping to determine how to 
 move forward with the DIP, we'd like to know if anyone is 
 actually using the feature and, if so, what your experience 
 with it has been.



One goal behind my work on Vibe.d and the vibe-http module is to 
increase  nogc
compatibility. Exceptions are widely used in Vibe.d and this is 
one of the
deal-breakers against this goal.
I haven't actually used  nogc exceptions yet, but it would 
definitely help
projects such as Vibe.d that are trying to give users control 
over memory
management and allocation strategies, without having to rewrite 
existing
exception handlers using e.g. `stdx.allocator`.

As an example, the HPACK library that is currently being used in 
vibe's new http2
module would be fully  nogc compatible if it wasn't for 
exceptions. See these
design constraints:

https://github.com/vibe-d/vibe-http/wiki

If there's some interest on resuming this DIP I would be happy to 
start testing
 nogc exceptions in my modules.
Mar 06 2019
prev sibling next sibling parent Atila Neves <atila.neves gmail.com> writes:
On Wednesday, 6 March 2019 at 12:36:56 UTC, Mike Parker wrote:

 Some time ago, Walter asked me to postpone DIP 1008, 
 "Exceptions and  nogc". Subsequently, an implementation was 
 merged behind the -dip1008 flag. In helping to determine how to 
 move forward with the DIP, we'd like to know if anyone is 
 actually using the feature


Quite a bit.


 and, if so, what your experience with it has been.


Well, it doesn't work, as mentioned above. But I've come to use 
it anyway because I'm ok with "this code is nearly  nogc, the 
exception (hah) being exceptions".

If it worked (especially the part regarding calling destructors 
on caught RC exceptions) I'd've already announced this:

http://code.dlang.org/packages/nogc

But it doesn't, so I haven't.

AFAIC all it needs are for the bugs to be fixed.
Mar 06 2019
prev sibling parent Peter Jacobs <peterj mech.uq.edu.au> writes:
On Wednesday, 6 March 2019 at 12:36:56 UTC, Mike Parker wrote:

 Some time ago, Walter asked me to postpone DIP 1008, 
 "Exceptions and  nogc". Subsequently, an implementation was 
 merged behind the -dip1008 flag. In helping to determine how to 
 move forward with the DIP, we'd like to know if anyone is 
 actually using the feature and, if so, what your experience 
 with it has been.


We use it in our compressible flow solver 
(http://cfcfd.mechmining.uq.edu.au/eilmer/).  It is a working 
research code.  We have lots of  nogc functions in the core of 
our solver but really want to be able to throw exceptions.  So we 
throw exceptions with simple messages in the optimized build and, 
when we want more information, we run the simulation again with a 
debug build of the code that constructs more informative messages 
in debug{} blocks.
Mar 07 2019