October 2006 - Cvs-nhc98

patch applied (hat): Add new command-line flag -iirename for debugging the compiler.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu Aug 23 07:59:07 PDT 2001 malcolm * Add new command-line flag -iirename for debugging the compiler. Shows the intermediate renaming tables between each import. Merged from stable-1-06-branch. M ./src/compiler98/Flags.hs -3 +7 M ./src/hattrans/Flags.hs -3 +7

1 0

patch applied (hat): Rename TokenId.qualImpRev -> qualImpPrel
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu Aug 23 07:01:14 PDT 2001 malcolm * Rename TokenId.qualImpRev -> qualImpPrel and TokenId.visImpRev -> visImport M ./src/compiler98/DbgId.hs -20 +22 M ./src/compiler98/DbgReplacePrelude.hs -2 +2 M ./src/compiler98/DbgTrans.hs -2 +2 M ./src/compiler98/DeriveShow.hs -2 +2 M ./src/compiler98/TokenId.hs -148 +148

1 0

patch applied (hat): In the user guide, the name Hatobserve becomes hat-observe again.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu Aug 23 06:27:37 PDT 2001 malcolm * In the user guide, the name Hatobserve becomes hat-observe again. M ./docs/hatuser.tex -12 +12

1 0

patch applied (hat): Undo a change in combined import qualified/unqualified behaviour which
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu Aug 23 06:22:33 PDT 2001 malcolm * Undo a change in combined import qualified/unqualified behaviour which slipped in with the last commit. M ./src/compiler98/IExtract.hs -3 +3

1 0

patch applied (hat): Minor formatting/comment changes.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu Aug 23 04:11:06 PDT 2001 malcolm * Minor formatting/comment changes. M ./src/compiler98/Bind.hs -2 +1 M ./src/compiler98/IExtract.hs -3 +5 M ./src/compiler98/Import.hs -19 +1 M ./src/compiler98/Overlap.hs -1 +1 M ./src/compiler98/Rename.hs -1 +2

1 0

patch applied (hat): Insert casts to get rid of C compiler warnings about " assignment from
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri Aug 17 08:40:51 PDT 2001 malcolm * Insert casts to get rid of C compiler warnings about "assignment from an incompatible pointer type." M ./src/hattools/observeutils.c -3 +3

1 0

patch applied (hat): Fix for Mark Tehver' s bug report.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri Aug 17 07:04:15 PDT 2001 malcolm * Fix for Mark Tehver's bug report. module Test2(Test(..)) where data Test {a0::Int, b0::Int} module Test1(fn) where import Test2 fn (Test {a0=a}) = a0 t + a gave "Unbound Identifier a0" error. The error was only visible with the combination of three circumstances: * field name is imported * field name is used in a lhs pattern * field name is also used as a selector on the rhs of the *same* equation. The fault was that the need-analysis was binding the selector name as if it were a pattern variable for the equation (as with puns), rather than generating a need for the imported selector. M ./src/compiler98/Need.hs -1 +1

1 0

patch applied (hat): Fix final problem for hat-observe:
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu Aug 16 07:11:06 PDT 2001 thorsten * Fix final problem for hat-observe: applications may indeed reference name nodes, which are somewhere forward in the Redex Trail file, as SAT nodes may point forward. So, the argumentation saying "it's enough to make a linear scan through a file" to find all concerning application nodes is finally disproved. New solution implemented: - use same approach as before, as long as function references point backwards (fortunately almost always the case) - if the function reference of an application eventually references a forward node (very rare case), then follow the entire reference to find out, whether it's an application of interest or not. M ./src/hattools/hat-observe.c +1 M ./src/hattools/observeutils.c -3 +8

1 0

patch applied (hat): Change internal representation of export decls.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Wed Aug 8 07:58:19 PDT 2001 malcolm * Change internal representation of export decls. * Previously, the two cases module M where module M () where were not distinguished! The former should export everything, the latter nothing. * In the old representation both cases were [] :: [Export id]. * In the new representation :: Maybe [Export id], Nothing means nothing, and Just [] means everything. (Just [...] gives an explicit list of export decls much as before.) M ./src/compiler98/AuxFile.hs -6 +19 M ./src/compiler98/AuxLabelAST.hs -3 +5 M ./src/compiler98/Need.hs -1 +3 M ./src/compiler98/Parse2.hs -3 +3 M ./src/compiler98/PreImport.hs -2 +7 M ./src/compiler98/PrettySyntax.hs -6 +4 M ./src/compiler98/Syntax.hs -1 +1 M ./src/compiler98/TraceTrans.hs -2 +6 M ./src/hattrans/AuxFile.hs -6 +19 M ./src/hattrans/AuxLabelAST.hs -3 +5 M ./src/hattrans/TraceTrans.hs -2 +6

1 0

patch applied (hat): - modification to hatinterface.c:
by Malcolm Wallace 10 Oct '06

10 Oct '06

Wed Aug 1 07:18:01 PDT 2001 thorsten * - modification to hatinterface.c: Apply further (and final) restrictions for the detection of application results. Now, function getResult should detect all cases of nodes, which do not have a result. - observe.c: Algorithm now finds really all (hopefully) applications of the given identifier. Even recursive applications, refering to a node somewhere forward in the trace file are now found. Problems with partial applications have now all (?) been solved, see correction of the "getResult" function. - detect.c: Enhanced algorithm to allow recursive/partial applications to be recognised correctly. They are now to be found at the correct position within the EDT. - hat-detect.c: Redex Trail browser (hat-trail) may now be launched from hat-detect. By answering a question with "r", the java tool is launched and the rhs or lhs of the equation shown in hat-detect can be traced. M ./src/hattools/detect.c -19 +64 M ./src/hattools/detect.h -1 +1 M ./src/hattools/hat-checki.c -4 +10 M ./src/hattools/hat-detect.c -32 +20 M ./src/hattools/hat-observe.c -28 +64 M ./src/hattools/hatinterface.c -7 +60 M ./src/hattools/hatinterface.h -1 +4 M ./src/hattools/observeutils.c -19 +166 M ./src/hattools/observeutils.h +11

1 0