import an LOOP macro

[masinter]

No description provided.

[pamoroso]

On Medley Online, when I load CML-LOOP.DFASL I get this break window with the error Can't find package for DEFINE-FILE-INFO reader environment XLP:

Not sure I'm loading it correctly.

[masinter]

Tbh I don’t know how to set up the package declaration so it works with makefile.--LarryMasinter.netOn Mar 14, 2024, at 9:19 AM, Paolo Amoroso ***@***.***> wrote: On Medley Online, when I load CML-LOOP.DFASL I get this break window with the error Can't find package for DEFINE-FILE-INFO reader environment XLP: cml-loop-error.png (view on web) Not sure I'm loading it correctly. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>

[pamoroso]

This and the documentation it references might help setting up the package for use with the File Manager.

[masinter]

I changed the file name to XCL-LOOP and added it to the loadup.
I fixed the package problems I hope.

[pamoroso]

Now it seems to be working. Pretty cool.

[nbriggs]

Why XCL- instead of CML?

Should loop be in its own package or should it be part of cl:/lisp: package if it's part of the language spec (or xcl: if it's "our own" implementation)

[MattHeffron]

I concur with @nbriggs .
Why is this in the "XLP" package? Doesn't that mean that all of the symbols that the loop macro uses for its extended form would have to be imported into whatever package the code using it is read in? Or all used as xlp::symbol?
In fact, the loop macro itself is in the "XLP" package and it isn't even exported from that package. So, even to use the simplest loop form requires either:
(xlp::loop forms)
or to (import (find-symbol "loop" "xlp") "name of package of code using loop") before the code using the loop macro is read,
or to put a similar import as part of the DEFPACKAGE of the file environment of a file using the macro.

If it's part of the standard, then put it into the correct package.

Is the point not to redefine the existing loop macro as defined in CMLSPECIALFORMS? It looks like that implementation and syntax would be handled correctly by the new fuller loop macro.

[masinter]

Interestingly the other words in the extended f op rm are pname only and can be in any or no (I interned at all. Only two symbols are exported: loop and loop-finish. That's why it works without the contortions @MattHeffron predicted. Try it!

[pamoroso]

The latest version seems to be working fine and no special handling of packages is required. Here's a quick test of evaluating at a XCL Exec a few loop expressions from Peter Seibel's book:

Running an infinite loop with (loop) works too.

[pamoroso]

A call to this implementation of loop to iterate over a hash table produces the error Undefined car of form XLP::WITH-HASH-TABLE-ITERATOR as reported in issue #609.

[nbriggs]

Both WITH-HASH-TABLE-ITERATOR and WITH-PACKAGE-ITERATOR definitions seem to be missing. (Latest version of the branch moves the package for the internal mechanism of the loop macro to LOOP: and sorts out the conflict with CMLSPECIALFORMS which already defined LOOP)

[nbriggs]

The author of the loop.lisp has the hash table bits as well.

[masinter]

I changed the package to "LOOP" and made sure "LOOP" and "LOOP-FINISH" were exported from the LISP package.

[masinter]

i spent the better part of a day on Both WITH-HASH-TABLE-ITERATOR and WITH-PACKAGE-ITERATOR and have decided to put that on ice because I don't want to change Medley's implementation of hash-tables or packages.
Those two features aren't part of LOOP.
I did find a couple of differences/bug fixes with a newer loop.liap.

[masinter]

I missed the parts about -TABLE-ITERATOR and WITH-PACKAGE-ITERATOR being used in some LOOP iterators.
But I don't want to hold this for them. I made a separate issue.