Session chair until Tu 15:40: GW
Session chair from Wed 18:00-19:30 AE
Andre Haley has missed this meeting.
Nothing of note.
Nothing of note.
Continues informally.
Not as much closed as we wanted.
Nothing of note
Was not present at the time
Nothing of note
Nothing of note
All elected officers accept the election
Ulrich Hoffman volunteers 11Y:0N:0A
Peter Knaggs volunteers 11Y:0N:1A
Gerald Wodni volunteers 11Y:0N:0A
Bernd Paysan volunteers 11Y:0N:0A
We make a snapshot and will review the document for the next meeting.
The consensus from 2020 has evaporated, leading to a long discussion. Workshop set up for a potential alternative proposal, revisit in main session on Wednesday.
After a workshop and some more discussions, a consensus emerged for the following new text:
A value that can be passed to EXECUTE (6.1.1370)
As a result, AE posted a new version of the proposal and this was voted on with vote #21: 12Y:0:0.
Presentation by Bernd Paysan. Discussion. Several participants found it too minimal. Some criticism of the factoring. Workshop proposed.
See workshop report below, which also covered the next two topics.
Some discussion. Some people think it goes too far. At the same time, some consider it too inflexible.
Several people like the terms proposed in the proposal. Discussion on whether “perceptor” is actually needed.
Some discussion of whether every word should be tickable. SP volunteers to write a proposal.
Are there any systems where EMIT does not send raw bytes? Find out with a CfV.
Go to committee vote. AE. Vote #22.
Change the proposed sentence as follows:
Append the semantics identified by xt to the execution semantics of the current definition.
Committee vote on that. AE. Vote #23: 12Y:0:0
Discussion about how to deal with the test suite in the document and forth-standard.org.
Contact Gerry Jackson on working together for making it easier to get the code from the test suite into forth-standard.org. GW
Either add some markup to the suite to allow integrating it or do some more automatic program slicing.
Some discussion about the name of +PLACE and the wording.
UH updates the proposal for the suggested wording and prepares a committee vote.
Do we want to be able to write a user-defined text interpreter, and do we want to be able to do it with FIND?
Retract. AE Done.
Contact ruv about this proposal. Action: AE
Contact ruv about this proposal. Action: UH
A system is allowed to do it, but is not required to do it. The committee does not see a need to add such a clause to the standard. AE. Done
Throw -13 is name is not found.
performing the semantics identified by xt
and then let the community have a go at it until the next meeting
SP writes up an answer and closes it. Done.
Answer by BP and closed. Done.
GW answers it. Done
Write an answer AE. Done
SP closes it again. Done
Closed by AE. Done.
Answer on behalf of the committee and close. AE. Done
Answer on behalf of the committee and close. AE. Done
Closed by AE. Done.
BLOCK
transfer from mass storage in the case when block u is already in a block buffer? (block, BLOCK #179)Closed by AE. Done.
Answer by GW on behalf of the committe. Done
Answer by GW on behalf of the committee. Done
Answer by UH on behalf of the committee. Action: UH
Leave it open. github issue to accept reference implementations. AE. Done
Leave it open. It will be accepted as reference implementation in a future version.
Reply and close by AE. Done
Ruv should run it through the test suite. Action: ??
Leave it open. Update the document with AE’s implementation. Action: PK
Response. AE. Done
Reply and closed by AE. Done
Leave it open. [COMPILE] is obsolescent.
AE and PK replied.
Action: PK
Closed by AE. Done
Closed by AE. Done. However, the editor might want to look at AE’s answer from 2021-01-14. Action: PK
Editing fix. Action: PK
Answered by AE. Done
Closed by UH. Done
Closed by AE. Done
Investigate the error in converting the document. Action: PK; Action: GW
Answer and closed by AE. Done
Closed by SP: Done
Maybe change 3.1.4. Possible Action: PK (Looking at 3.1.4 again, I don’t see how it could be improved)
Leave it open
Invite MitraArdron to collaborate. UH. Done
Closed by AE. Done.
Editing fix. Action: PK
Report by UH
We discussed a blend of the Minimal RECOGNIZER API proposal (Bernd), Recognizer Sequences (Anton) and the Recognizer Rewording Proposal (uho)
Just before and at the beginning of our recognizer workshop, I (uho) implemented the minimal proposal for seedForth. RECOGNIZER CORE (FORTH-RECOGNIZER to be named differently?, NOTFOUND) and seedForth-specific recognizers (for words, numbers). That took about 45 minutes and some debugging.
Recognizers have the stack effect ( addr u – i*x rectype | notfound )
For RECOGNIZER EXT we agreed on having:
the RECOGNIZER: ( xt-interp xt-comp xt-post <name> -- ) defining word
the COMPOUND-RECOGNIZER: ( rec1 rec2 ... recn n <name> -- ) defining word
the word GET-RECOGNIZERS ( compound-rec -- rec1 rec2 ... recn n ) to retrieve the recognizers that make up a compound recognizer
the word SET-SECOGNIZERS ( rec1 rec2 ... recn n compound-rec -- ) to store the recognizers in a compound recognizer
the lit-recognizer
All names are subject to change.
Given the two variations to handle STATE (either in RECOGNIZER:’s DOES> part or in INTERPRET), yesterdays participants favoured to have the single occurrence of STATE in INTERPRET. Further investigation and model implementations will show whether one or the other is beneficial.
Resulted in the text for the proposal Reword the term “execution token”
Report by GW
The workflow is as follows:
make pdf
Stack comments should be parseable for automatic processing
Meeting in half a year, for an afternoon. Agenda for that amount of time. Action: UH.
Thanks to SP for serving as chair
Half-year meeting 2022-02-18 13:00 UTC Full meeting in front of EuroForth, probably mid-September 2022