| |
![deadalnix's Gravatar profile Gravatar of deadalnix](//www.gravatar.com/avatar/0bc624bb19867fc4ff397ffec40b1adf?d=identicon&s=32) | Posted by deadalnix in reply to russhy | Permalink Reply |
|
deadalnix ![deadalnix's Gravatar profile Gravatar of deadalnix](//www.gravatar.com/avatar/0bc624bb19867fc4ff397ffec40b1adf?d=identicon&s=80)
Posted in reply to russhy
| On Friday, 15 October 2021 at 13:21:25 UTC, russhy wrote:
> On Friday, 15 October 2021 at 11:57:28 UTC, deadalnix wrote:
> On Friday, 15 October 2021 at 10:58:14 UTC, bauss wrote:
> I think that's his problem. D doesn't have a specific philosophy about memory management.
This is fine. But here is the deal: that means D cannot have a philosophy on safety. Or have closure allocated on the GC by default. Or a gazilion other things.
It's the same problem, just in another dimension: a decision is made, ut then consequences are not accepted.
using allocators aware APIs solves all of this, look at zig!
it is a problem already solved, use allocators and don't assume one is using GC or other means
Many D construct assume a GC. Which kinda is the point. Either you don't assume a GC and go the zig road, or you assume a GC. But D is trying to have its cake and eat it too, and the result is schizophrenic in nature.
At some point you need to chose a path. And if you don't want to chose a path at level N to leave that to the user, then you need to pick a path at level N-1 or bellow.
|