More operator tables #86
Conversation
rocky
force-pushed
the
more-operator-tables
branch
from
ed2c240 to
7469472
Compare
mathics_scanner/data/operators.yml
Outdated
| @@ -5284,7 +5311,7 @@ PlusMinus: | |||
|
|
|||
| Postfix: | |||
| actual-precedence: 640 | |||
| precedence: 640 | |||
| precedence: 70 # Seems weird but this is what old Mathics code has | |||
There was a problem hiding this comment.
In WL it is 70, so I think it is correct
There was a problem hiding this comment.
I will remove the comment then. Thanks!
mathics_scanner/data/operators.yml
Outdated
| @@ -7839,7 +7866,7 @@ Xnor: | |||
|
|
|||
| Xor: | |||
| actual-precedence: 280 | |||
| precedence: 215 | |||
| precedence: 220 | |||
There was a problem hiding this comment.
I will make note of this for the future in a comment FIXME.
The change to Mathics core which is coming up will be a bit large and probably painful for me. I will probably have to make several passes to get everything tidy.
But initially, I am going to be conservative and go with existing Mathics code. After things pass I can go back and fix up the list of discrepancies that were found.
mathics_scanner/data/operators.yml
Outdated
| meaningful: true | ||
| # comments: | ||
|
|
||
| UndirectedEdge: | ||
| actual-precedence: 370 | ||
| precedence: 295 | ||
| precedence: 120 |
There was a problem hiding this comment.
Thanks again for checking. I will again make a TODO fixme for this.
Some will have to be FIXME's until we get everything merged in and we can start a 2nd round of testing/adjusting.
| @@ -168,15 +177,15 @@ AngleBracket: | |||
| Apply: | |||
| actual-precedence: 820 | |||
| Precedence-Function: 620 | |||
| precedence: 626 | |||
| precedence: 620 | |||
There was a problem hiding this comment.
I guess the criterium now is what Precedece[] says, isn't it?
There was a problem hiding this comment.
I am probably not using the terms in the way that Robert Jacobson had intended.
Right now, "precedence" is what mathics-core uses as precedence, whether it is Precendence[] or not.
Ideally, this would be the same as Precedence[] and to the extent we can make it that we should. Only in those cases where we find we can't, I envision the Precedence-Function to give the Precedence[] value. And where they are the same there is no need to include this field in the table duplicating the "precedence" value.
Down the line, the "actual-precedence" field can be removed.
However, right now I want to get something minimally working. Unfortunately, just this is already a big change. But a really good one! We will be much closer to providing everything the way it appears in WMA with reduced hand-coded values strewn about in Python code.
As I type this I have finally gotten Mathics-core to pass the regular tests using data from this YAML file. The manifest stuff will require more work.
Obviously this code has to get merged before mathics-core. So that will happen soon.
|
LGTM. @rocky, merge this when you feel it is ready. |
This can now be used with upcoming code to Mathics-core which will replace some of the operator-coded information in Python files.
However much more would be nice.
|
BTW - what remains in this repository is to add more of the operator symbols to the scanner in a way keyed from the JSON file. This will be coming up next. |
Most of the information we will need in mathics-core to revise
mathics.core.parser.operators.