Re: sb-thread:make-thread sets thread's package to CL-USER

On Wed, Jun 12, 2024 at 5:42 PM Stas Boukarev <[email protected]> wrote:

> It's not the REPL, it's LOAD.
>

Right, it is behaving like this when source files are LOAD'ed , not sure
about REPL.
Anyway, once I know this is the intended behaviour I can set the needed
package explicitly
in thread code – thanks for pointing to the explanation.



>
> On Thu, Jun 13, 2024 at 12:40 AM Michał "phoe" Herda via Sbcl-help <
> [email protected]> wrote:
>
>> Yes. It's possible that your REPL bound *PACKAGE* somewhere along the way
>> and that this is the binding you are seeing and modifying; this binding is
>> effectively thread-local. If you want to copy the value of that dynamic
>> binding to another thread, use something along the line of:
>>
>> (let ((package *package*)) (sb-thread:make-thread (lambda () (let
>> ((*package* package)) ...))))
>> On 12.06.2024 23:28, Raimondas K. wrote:
>>
>> (sb-ext:symbol-global-value '*package*) both in parent thread
>> and new thread return #<PACKAGE "COMMON-LISP-USER">
>>
>> Thank you for the pointer to the manual – I think, this explains what I
>> am seeing:
>> "threads do not inherit dynamic bindings from the parent thread"
>>
>>
>>
>> On Wed, Jun 12, 2024 at 5:14 PM Michał "phoe" Herda via Sbcl-help <
>> [email protected]> wrote:
>>
>>> What does (sb-ext:symbol-global-value '*package*) return in your REPL?
>>> Is it consistent with what
>>> https://www.sbcl.org/manual/#Special-Variables says?
>>> On 12.06.2024 22:58, Raimondas K. wrote:
>>>
>>> The *package* value in the new thread is not NIL and not what was in the
>>> calling thread,
>>> so the question whether the binding was not retained or set to an
>>> unexpected value
>>> is not as interesting as what the behaviour should be, logically.
>>>
>>>
>>> On Wed, Jun 12, 2024 at 4:41 PM Stas Boukarev <[email protected]>
>>> wrote:
>>>
>>>> It doesn't set anything, it just doesn't retain your binding of
>>>> *pacakge* at the time of calling make-thread.
>>>>
>>>> On Wed, Jun 12, 2024 at 11:37 PM Raimondas K. <[email protected]> wrote:
>>>>
>>>>> I don't see this explicitly documented either way,
>>>>> but it looks that, empirically, the *package* inside the
>>>>> newly created thread is set to COMMON-LISP-USER.
>>>>>
>>>>> To demonstrate
>>>>> (defpackage :my (:use :cl))
>>>>> (in-package :my)
>>>>>
>>>>> (defun test ()
>>>>>   (format *standard-output* "in TEST thread: package ~a~%" *package*))
>>>>>
>>>>> (format *standard-output* "in MAIN thread: package ~a~%" *package*)
>>>>>
>>>>> (sb-thread:make-thread #'test
>>>>>                          :arguments '()
>>>>>                          :name "test")
>>>>>
>>>>> SBCL 2.3.2 prints out
>>>>> in MAIN thread: package #<PACKAGE "MY">
>>>>> in TEST thread: package #<PACKAGE "COMMON-LISP-USER">
>>>>>
>>>>> Not sure if this is how it is supposed to work; I find this
>>>>> behaviour counterintuitive, c/w for instance, with (load "file").
>>>>>
>>>>>    =R.
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Sbcl-help mailing list
>>>>> [email protected]
>>>>> https://lists.sourceforge.net/lists/listinfo/sbcl-help
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> Sbcl-help mailing [email protected]://lists.sourceforge.net/lists/listinfo/sbcl-help
>>>
>>> _______________________________________________
>>> Sbcl-help mailing list
>>> [email protected]
>>> https://lists.sourceforge.net/lists/listinfo/sbcl-help
>>>
>> _______________________________________________
>> Sbcl-help mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/sbcl-help
>>
>

_______________________________________________
Sbcl-help mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sbcl-help