The GC series[1] on D blog is great, but there’s still one piece missing: heap-allocated class instances.

[1] https://dlang.org/blog/the-gc-series/

On Thursday, 2 July 2020 at 22:03:46 UTC, Ogi wrote:
> The GC series[1] on D blog is great, but there’s still one piece missing: heap-allocated class instances.
>
> [1] https://dlang.org/blog/the-gc-series/

I held off for a long time because I found the conflation of finalization and destruction embarrassing to advertise, but leaving it out of the article would have been wrong. Now that we've got GC.inFinalizer, I can move forward. It's just a matter of making the time for it. I have so many things going on I keep putting it off, but it won't be much longer.

On Friday, 3 July 2020 at 03:39:52 UTC, Mike Parker wrote:
> On Thursday, 2 July 2020 at 22:03:46 UTC, Ogi wrote:
>> The GC series[1] on D blog is great, but there’s still one piece missing: heap-allocated class instances.
>>
>> [1] https://dlang.org/blog/the-gc-series/
>
> I held off for a long time because I found the conflation of finalization and destruction embarrassing to advertise, but leaving it out of the article would have been wrong. Now that we've got GC.inFinalizer, I can move forward. It's just a matter of making the time for it. I have so many things going on I keep putting it off, but it won't be much longer.

Sometimes making people aware of the dark side is also important. So a blog post should strike a balance between the two...more transparency.

But I guess in your you kind of were waiting for the bus GC.inFinalizer

On Friday, 3 July 2020 at 09:19:27 UTC, aberba wrote:

>
> Sometimes making people aware of the dark side is also important. So a blog post should strike a balance between the two...more transparency.

Agreed! If you look through the blog history, you'll find I don't shy away from it generally. That said, the purpose of the blog has evolved since its inception. As the *official* D blog, it's now a major marketing tool, so we do have to be more careful about tone than we were in the first year or so before we thought of it that way. That wasn't the consideration in this case, though. I expected the issue would be resolved sooner or later, so I thought it better to wait.

>
> But I guess in your you kind of were waiting for the bus GC.inFinalizer

Yes. I was hoping more for the separation of finalization and destruction. I started a thread about it here a while back and have discussed it with others at DConf and via email, and I have thought about different approaches for implementing it in the hopes of recruiting someone to write a DIP for it (writing DIPs is not something I want to do as DIP manager). The `inFinalizer` thing gets the job done, though, and without breakage.

On Friday, 3 July 2020 at 09:52:54 UTC, Mike Parker wrote:
> in the hopes of recruiting someone to write a DIP for it (writing DIPs is not something I want to do as DIP manager).

On the other hand, I think you could also consider the reverse: recruit an alternative DIP manager who will manage the status of DIP/DIPs you write.

On Friday, 3 July 2020 at 03:39:52 UTC, Mike Parker wrote:
> I held off for a long time because I found the conflation of finalization and destruction embarrassing to advertise, but leaving it out of the article would have been wrong. Now that we've got GC.inFinalizer, I can move forward. It's just a matter of making the time for it. I have so many things going on I keep putting it off, but it won't be much longer.

BTW, I’ve translated the series so far into Russian [1].

[1] https://habr.com/post/509316/