Add support for ABCL and Clasp #2
Conversation
micros.asd
Outdated
| @@ -1,41 +1,48 @@ | |||
| (defsystem "micros" | |||
| :depends-on () | |||
| :depends-on (:bordeaux-threads) | |||
There was a problem hiding this comment.
Adding a dependency in micros may risk a conflict between the library used by the user and the library used by micros, causing a malfunction.
Is there any way to avoid this?
There was a problem hiding this comment.
Ummm, I don't know really, if we want to make this compiled alone (without the Lem system), I think it's necessary
There was a problem hiding this comment.
For example, if users are developing bordeaux-threads, changing their implementation will likely cause micros to stop working.
There was a problem hiding this comment.
oooh, I see ummm, then maybe use some specific functions depending on the implementation, a little more manual, but that can be a solution
There was a problem hiding this comment.
Ok, so.... SBCL implement a couple of more things for hash-tables to work concurrently so I'm experimenting with https://github.com/no-defun-allowed/concurrent-hash-tables .
It doesn't solve the problem, given that we still have to use external dependencies, so I think if this version works for more implementation, the solution might be to fork it and integrate it in micro as a internal system.
There was a problem hiding this comment.
It's a pain if there is a but, because we would have to go to the main repo to check it, buut I'll add the default option for default SBCL to work with the native calls, so at least it works as intended in the reference implementation.
There was a problem hiding this comment.
The concurrent-hash-table is very interesting, thanks for sharing.
In this case, we need a mutex of hash-table in eval-for-emacs, but I don't think it requires much performance.
What about going with a simple with-lock-held pattern?
There was a problem hiding this comment.
Still, I would have to use the bt library for that (if I want it to be portable)
There was a problem hiding this comment.
micros(swank) has definterface and defimplementation.
Since we already have the bare minimum for threads, why not just use them?
There was a problem hiding this comment.
So, seems like swank already have some thread-locks implementations 🤦 , the code seems fine but it doesn't seem to work for SBCL.
swank.lisp
Outdated
| ;; Needless to say, this also excludes some internal bits, but | ||
| ;; getting there is too much detail when "unspecified" says what | ||
| ;; is important -- unpredictable, but harmless. | ||
| `(bt:with-recursive-lock-held((hash-table-lock ,hash-table)) |
There was a problem hiding this comment.
Is it ok if don't add sb-int for hash-table-count?
There was a problem hiding this comment.
Sorry, what do you mean?
There was a problem hiding this comment.
Ah, it was hash-table-lock, not hash-table-count.
In other words, I think the function is unbound.
(fboundp 'hash-table-lock) => NIL
There was a problem hiding this comment.
Ups! Sorry, I was looking at the implementation and seems very sbcl specific
(https://www.sbcl.org/manual/#Hash-Table-Extensions)
There was a problem hiding this comment.
I don't really know how to solve it easily
There was a problem hiding this comment.
As a workaround, what about using mutex for except sbcl?
There was a problem hiding this comment.
I'll take a look and see if it works, thanks 👍
There was a problem hiding this comment.
Related to the top comment
|
Thank you so much 🙏 |
Adapt the SLIME backed for Micros, it has been superficially testes tho, so there is probably need for some more testing.