While I'm just trying to figure out the language at this point, you could try something involving praxi.
On Saturday, June 26, 2021 at 4:52:24 PM UTC d4v3y_5c0n3s wrote: > Sorry for the delay on this response. > > Basically, as for having ATS thing that a pointer is valid, you'd > create an "extern" function like this: > extern fn name_here () : [l:addr] (typename @ l | ptr l)// where this > function points to something in C > > As for not having to "psuedo-free" the return value every time you are > done with it, there are several things that can be done. First, in certain > instances you can use templates to make things easier to manage, but > explaining exactly what I mean is kind of difficult to articulate over > Google Groups. You'll have to take my word for it, and you'll likely > understand when you play with templates more. > Second, this "psuedo-free" issue is a natural consequence of using > linear types, these restrictions are designed to make resources less likely > to be mismanaged. Rather than trying to avoid this pattern completely, I'd > suggest formalizing this relationship with ATS' type system. I'd take a > look at the "Reader and Writer" pattern here to get an idea of what I'm > suggesting. Basically, you'd have a view for the memory IO that acts as a > "key" of sorts. Whenever you call the function I suggested above, this > view would be consumed, preventing anyone else from calling that function. > When you are done, you'd call another function that consumes this pointer > to the memory IO and gives you back the "key" view. Now, you can call the > first function again since you've finished touching that memory. The > benefit of doing this is that wouldn't have to worry about things like > calling the function at the same time in concurrent code. It also looks > better, so there's that. > > On Thursday, June 3, 2021 at 6:35:40 PM UTC-4 d4v3y_5c0n3s wrote: > >> I'd suggest using an external function/value with dependent types that >> tell ATS the "sort" of the pointer is "agz" (arguements greater than >> zero.) I'll try to put together an example. >> >> On Thu, Jun 3, 2021, 6:15 PM David Smith <sodab...@gmail.com> wrote: >> >>> >>> When dealing with memory mapped IO, is there a neat way to axiomatically >>> say "there is something at address 0x12345678? >>> >>> At the moment I'm using a C macro that just returns the address, and >>> then have an at-view in the ATS type signature, though I'd prefer to >>> somehow treat that as global, if only so I don't have to "pseudo-free" the >>> stuff every time I use it. >>> >>> Thanks in advance. >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "ats-lang-users" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to ats-lang-user...@googlegroups.com. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/ats-lang-users/baa3dbd2-410f-44d5-be2e-2bbd6e7efcf5n%40googlegroups.com >>> >>> <https://groups.google.com/d/msgid/ats-lang-users/baa3dbd2-410f-44d5-be2e-2bbd6e7efcf5n%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- You received this message because you are subscribed to the Google Groups "ats-lang-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to ats-lang-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/c96724af-3cdc-4e2a-a501-5aff4de0c2a9n%40googlegroups.com.
