Add pattern names in SyntaxInformation and missing usage, syntax information and syntax autocompletion

[maxitg]

Changes

• In the Mathematica options inspector, there is an option called "HighlightMissingArgumentsWithTemplate".
• If you turn it on, in case of missing arguments, instead of displaying a single red arrow, you will get a helpful hint with argument names which can serve as a super fast lightweight documentation (like you would get in, e.g., Xcode after autocompleting a function):

• The names of the arguments are taken from pattern names in SyntaxInformation.

• This PR adds names to all SyntaxInformation except for the built-in RulePlot thus enabling this feature for all SetReplace symbols.

• This also adds missing usage messages, syntax information, and syntax autocompletion.

• Expands argument names in usage messages to be descriptive so that using autocompleted templates are more informative.

Comments

• Null is due to OptionsPattern[]. It can be changed to, say, opts if that pattern is named as opts : OptionsPattern[]. However, if one does that, some optional arguments will always be displayed in the template, which is, of course, unacceptable. I think we'll have to wait until this weed is fixed upstream:

• Autocompletion is generally disabled for option names. The problem is that I'm not aware of a way to enable autocompletion for multiple arguments at a time. In order to solve this, we need to use symbols instead of strings for options (Use symbols instead of strings for option names #604), and we need to figure out how to autocomplete the values for enum-like options (Autocomplete enum-like options #605).

---

This change is 

[maxitg]

@daneelsan, is there any reason not to merge this now while we try to figure out how to do autocompletion for options?

[daneelsan]

@maxitg I missed the last changes you made. I'll review it now.

[daneelsan]

Reviewed 4 of 31 files at r3, 1 of 2 files at r4.
Reviewable status: 5 of 32 files reviewed, 8 unresolved discussions (waiting on @maxitg and @taliesinb)

---

Kernel/GeneralizedGridGraph.m, line 32 at r5 (raw file):

(* Implementation *)

GeneralizedGridGraph[args___] := ModuleScope[

To change args_ -> dimensionSpecs_, opts : OptionsPattern[] the CheckArgumentCount check should probably done in GeneralizedGridGraph:

GeneralizedGridGraph[args___] :=
  Null /; CheckArguments[GeneralizedGridGraph[args], {1, 1}] && False;

then

GeneralizedGridGraph[dimensionSpecs_, opts : OptionsPattern[]] := ModuleScope[...]

We might want to standardize how we do argument check for all our exposed functions.

---

Kernel/HypergraphToGraph.m, line 43 at r5 (raw file):

(* main *)
expr : HypergraphToGraph[
      hgraph_,

hypergraph_,

---

Kernel/IndexHypergraph.m, line 22 at r5 (raw file):

(* main *)
expr : IndexHypergraph[hgraph_, start : _ : 1] := ModuleScope[

IndexHypergraph[hypergraph_, startIndex_ : 1]

---

Kernel/IsomorphicHypergraphQ.m, line 23 at r5 (raw file):

(* main *)
expr : IsomorphicHypergraphQ[hgraph1_, hgraph2_] := ModuleScope[

IsomorphicHypergraphQ[hypergraph1_, hypergraph2_]

---

Kernel/RandomHypergraph.m, line 25 at r5 (raw file):

(* Main entry *)
expr : RandomHypergraph[sig_, max_ : Automatic] := ModuleScope[

RandomHypergraph[sizeSpec_, maxVertices_ : Automatic]

---

Kernel/Subhypergraph.m, line 39 at r5 (raw file):

(* main *)
expr : Subhypergraph[arg1_, arg2_] := With[{res = Catch[subhypergraph[HoldForm @ expr, arg1, arg2]]},

arg1_, arg2_ -> hypergraph_, vertexList_

---

Kernel/ToPatternRules.m, line 70 at r5 (raw file):

  toPatternRules[#, caller] & /@ rules;

ToPatternRules[rules_] := ModuleScope[

hypergraphRules_

---

Kernel/WolframModel.m, line 39 at r5 (raw file):

With[{properties = $newParameterlessProperties,
      stepSpecKeys = Values[$stepSpecKeys]},
  FE`Evaluate[FEPrivate`AddSpecialArgCompletion["WolframModel" -> {{"PatternRules"}, 0, stepSpecKeys, properties}]]];

Interesting. So the list of strings autocompletes not only at level 1...

[maxitg]

Reviewable status: 5 of 32 files reviewed, 8 unresolved discussions (waiting on @daneelsan and @taliesinb)

---

Kernel/GeneralizedGridGraph.m, line 32 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

To change args_ -> dimensionSpecs_, opts : OptionsPattern[] the CheckArgumentCount check should probably done in GeneralizedGridGraph:

GeneralizedGridGraph[args___] :=
  Null /; CheckArguments[GeneralizedGridGraph[args], {1, 1}] && False;

then

GeneralizedGridGraph[dimensionSpecs_, opts : OptionsPattern[]] := ModuleScope[...]

We might want to standardize how we do argument check for all our exposed functions.

Why would we want to check in GeneralizedGridGraph? I think it would be better to do it consistently with throw[Failure[...]]. I agree wrt consistency, and I think we should use the new messages system for it.

---

Kernel/HypergraphToGraph.m, line 43 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

hypergraph_,

Done.

---

Kernel/IndexHypergraph.m, line 22 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

IndexHypergraph[hypergraph_, startIndex_ : 1]

Done.

---

Kernel/IsomorphicHypergraphQ.m, line 23 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

IsomorphicHypergraphQ[hypergraph1_, hypergraph2_]

Done.

---

Kernel/RandomHypergraph.m, line 25 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

RandomHypergraph[sizeSpec_, maxVertices_ : Automatic]

Done.

---

Kernel/Subhypergraph.m, line 39 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

arg1_, arg2_ -> hypergraph_, vertexList_

Done.

---

Kernel/ToPatternRules.m, line 70 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

hypergraphRules_

Done.

---

Kernel/WolframModel.m, line 39 at r5 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

Interesting. So the list of strings autocompletes not only at level 1...

Yes, it does.

[daneelsan]

Reviewable status: 5 of 32 files reviewed, 9 unresolved discussions (waiting on @daneelsan, @maxitg, and @taliesinb)

---

Kernel/GeneralizedGridGraph.m, line 32 at r5 (raw file):

Previously, maxitg (Max Piskunov) wrote…

Why would we want to check in GeneralizedGridGraph? I think it would be better to do it consistently with throw[Failure[...]]. I agree wrt consistency, and I think we should use the new messages system for it.

Right, also

func[args___] /; !CheckArgumentCount[...] := throw[...]

makes more sense than

Func[args___] := Null /; CheckArgumentCount[...] && False

---

Kernel/RandomHypergraph.m, line 31 at r6 (raw file):

(* Error messages *)
RandomHypergraph::invalidSizeSpec = "\

This message needs to be also updated also in Tests/.

[maxitg]

Reviewable status: 5 of 32 files reviewed, 9 unresolved discussions (waiting on @daneelsan, @maxitg, and @taliesinb)

---

Kernel/RandomHypergraph.m, line 31 at r6 (raw file):

Previously, daneelsan (Daniel Sanchez) wrote…

This message needs to be also updated also in Tests/.

I reverted it back to not break backwards compatibility unnecessarily.

[daneelsan]

Reviewable status: 5 of 32 files reviewed, 9 unresolved discussions (waiting on @daneelsan, @maxitg, and @taliesinb)