Commit Graph

17 Commits

Author SHA1 Message Date
Grigorii Kirgizov 119daf05d9 Restrict notion of a principal rule to not track unnecessary matches in journal
Now principal rule head must  include only principal constraints or be empty.
Without this restriction journal may include stale chunks which have no information to invalidate them.
2020-02-05 12:29:33 +03:00
Grigorii Kirgizov bc06bf39cc Handle one more case of finding principal constraints in Query rules with an additional heuristic. 2019-10-15 21:15:38 +03:00
Fedor Isakov 5258a9343f Switch to collecting types during typechecking using the explicit type collector.
Use the parameter to typechecking query TYPECHECK to collect types.
Stop highlighting every expression with its type.
2019-10-10 17:30:11 +02:00
Grigorii Kirgizov 2b1ec4a4a6 Remove interprocedural version of analysis 2019-08-22 15:16:18 +02:00
Grigorii Kirgizov 944a181685 Switch basic principals analysis to fully gen-time impl 2019-08-22 15:16:18 +02:00
Grigorii Kirgizov 9843669f79 Remove old runtime coderules analyses 2019-08-22 15:16:18 +02:00
Grigorii Kirgizov 37a6a4e1b7 Two more tests on FreeVarEscapeAnalysis for more complex cases. Extend checker logic for them. 2019-08-22 15:16:18 +02:00
Grigorii Kirgizov c6fc2c3f96 Test FreeVarEscapeAnalysis on full languages. Passes for BL, fails for lambdacalc as intended. 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov 1c9060b769 Add simpler analysis to answer a question "Is incrementality allowed for this Coderules aspect?". Add initial tests for it.
It is a rule-local, non-interprocedural analysis and it also uses VarAnalysisUtil.
New analysis checks whether free variables can "escape" principal constraint rules inside principal constraints.
If there is such escaped variable, then it is a signal that this type system is designed with non-local type inference in mind.
Otherwise there's no need to produce principal constraints with free logicals as arguments.
So, no principal constraints with free vars -- no type inference is possible.
2019-08-22 15:16:17 +02:00
Grigorii Kirgizov ae993f6868 Separate VarAnalysisUtil into base and impl part that handles impl of involved relations ("isAffected"). 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov 87fb26b95b Fix NPE in access to hashmap in VarModificationAnalysis 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov e41589cd4a Fix few places in VarModificationAnalysis. Add tests for it (found some problems). 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov f1aa5cb923 Finish interprocedural part of VarModificationAnalysis. Add one more test for Var..Util. 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov 4ab877e822 Refine isAffected relation in VarAnalysisUtil: handle dep-s through composite logicals 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov 62b08720cd Fix finding isAffected relation in var2notvar case by more carefully handling aliasing with "non-local" vars. Few more tests. 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov f7040b1d8d Add metalogical aliasing in VarAnalysisUtil. Rewrite older analyses fully in BL and not template macros. 2019-08-22 15:16:17 +02:00
Grigorii Kirgizov 8a2942c42b Move VarModificationAnalysisUtil to new model for analysis-related code 2019-08-22 15:16:17 +02:00