Move grammar generation to Style package #3662
Conversation
|
I probably don't have time to review this in the short term. |
M2/Macaulay2/packages/Style.m2
Outdated
| -- TODO: Move these two elsewhere: | ||
| Function and Function := (f, g) -> s -> f s and g s | ||
| Function or Function := (f, g) -> s -> f s or g s |
There was a problem hiding this comment.
Do you want to actually move these to Core while you're moving them? Maybe also add not Function and Function xor Function. Especially the former would be super useful for things like this:
i1 : not Function := f -> x -> not f x
i2 : select({0,1,0,2,0,3}, not zero)
o2 = {1, 2, 3}
M2/Macaulay2/packages/Style.m2
Outdated
| if #bad > 0 then error( | ||
| "encountered symbol(s) that are not alphanumeric or have less than 2 characters:", | ||
| concatenate apply(bad, (name, symb) -> { | ||
| "\n\t", | ||
| -* TODO: symbolLocation name, ": here is the first use of ", *- | ||
| toString name})); |
There was a problem hiding this comment.
I recently learned something super cool about how error prints symbols which I didn't know when I wrote this:
i7 : error splice("encountered symbol(s) that are not alphanumeric or have less than 2 characters:",
toSequence between_" " {symbol Module, symbol print})
stdio:13:5:(3): error: encountered symbol(s) that are not alphanumeric or have less than 2 characters:'Module' 'print'
../linuxbrew/.linuxbrew/share/Macaulay2/Core/expressions.m2:1234:42-1234:48: here is the first use of 'Module'
../linuxbrew/.linuxbrew/share/Macaulay2/Core/expressions.m2:1158:0-1158:5: here is the first use of 'print'There was a problem hiding this comment.
That's neat!
I'm a little confused at this whole error message, though. The symbols list appears to be made up of two things: the list of all packages and everything from the dictionaries in the pre-loaded packages. We already call select with okay on the latter, so we know that none of those will end up in bad. Why not just do the same thing on the former?
And also, these packages are always paired up with the Core symbol (I guess so that when we call is on them, we know that they're packages), so even if one of them ends up in bad and we modify the error message to print the symbol location, it will just be the location of Core.
There was a problem hiding this comment.
I'm not sure. Maybe experiment with not selecting on line 51 and see if anything breaks?
There was a problem hiding this comment.
There are tons of non-alphanumeric symbols (mathematical operators, all the Package$foo ones) that were getting filtered out by select.
The only elements of this list we weren't selecting out with okay are the package names, and we know that they'll be alphanumeric since newPackage raises an error when they're not. I just pushed a commit removing this check.
M2/Macaulay2/packages/Style.m2
Outdated
| cachedSymbols.Function = first \ select(symbols, isFunction); | ||
| cachedSymbols.Constant = sort join( | ||
| first \ select(symbols, isConst), | ||
| {"Node", "Item", "Example", "CannedExample", "Pre", "Code", "Tree", |
There was a problem hiding this comment.
Do you mind adding a comment close to this, and also somewhere in SimpleDoc, to keep these in sync?
There was a problem hiding this comment.
Is there a reason we hardcode these here instead of just exporting them from SimpleDoc and letting them get pickedup with all the other symbols? (I suppose Node is a class, so it would get a different color -- is that the main reason?)
There was a problem hiding this comment.
I just pushed a commit exporting these docstring symbols from SimpleDoc so we can stop hard-coding them here. I also renamed the Node class to DocNode so Node would still be the same color as the others.
d-torrance
force-pushed
the
generate-grammar
branch
from
eb50e2a to
df55263
Compare
Also add not and xor
d-torrance
force-pushed
the
generate-grammar
branch
from
df55263 to
4f0f9df
Compare
The "symbols" list is the union of two sets:
* {(pkgname, Core) : pkgname is the name of a package}
* {(name, symb) : key/value pair from a pre-loaded package dictionary}
pkgname will always be okay (we check for this in "newPackage", and we
already ran "select" w/ the okay function on the second set
M2/Macaulay2/packages/SimpleDoc.m2
Outdated
| "CannedExample", | ||
| "Code", | ||
| "Example", | ||
| "Item", | ||
| "Node", | ||
| "Pre", | ||
| "Synopsis", | ||
| "Tree", |
There was a problem hiding this comment.
I don't think these should be exported by SimpleDoc. Some of them are very common words and could lead to random frustrations.
There was a problem hiding this comment.
And case in point, we got the following build failure from the Varieties::tangentSheaf(ProjectiveVariety) example :)
i5 : Node = Proj(QQ[a,b,c]/ideal(b^2*c-a^2*(a+c)))
stdio:5:11:(3): error: assignment to protected global variable 'Node', originally defined at stdio:5:0:(3):I'll restore the original behavior but add some comments. The annoying this is that many of the SimpleDoc docstring symbols are in fact symbols defined in Core, but not these few.
d-torrance
force-pushed
the
generate-grammar
branch
from
ede5ae3 to
8676fd6
Compare
As suggested by @mahrud in #3298 (comment)), we move the
generateGrammarfunction from a build script to a method exported by theStylepackagee.