- Cvs-hugs - Haskell.org

cvs commit: hugs98/lib Prelude.hs hugs98/lib/exts ConcBase.hs
by Alastair Reid 03 Apr '01

03 Apr '01

reid 2001/04/03 12:48:49 PDT Modified files: lib Prelude.hs lib/exts ConcBase.hs Log: Never, ever, make a last minute cleanup as you're about to commit and fail to test it! :-) Moved blockIO from ConceBase back into Prelude. The alternative is to export toObj/fromObj (unsafe coercions) from Prelude - not attractive. Revision Changes Path 1.9 +7 -1 hugs98/lib/Prelude.hs 1.5 +0 -6 hugs98/lib/exts/ConcBase.hs

1 0

cvs commit: hugs98/lib Prelude.hs hugs98/lib/exts ConcBase.hs
by Alastair Reid 03 Apr '01

03 Apr '01

reid 2001/04/03 12:29:27 PDT Modified files: lib Prelude.hs lib/exts ConcBase.hs Log: When a thread blocks on an MVar, we have to take the complete state of the thread and store it in the MVar. This used to be easy because the state was all in the success continuation but exception handlers live on the stack so we have to unwind the stack making a list of all the exception handlers and store that list of exception handlers into the MVar too. This is done by returning a Hugs_BlockThread object. When catch sees one of these, it adds a Hugs_Catch into the object and returns Hugs_BlockThread again. Eventually, this gets up to the top thread loop which, finally, does the write into the MVar. Trivial changes: o Improved error messages if a forked thread ends in error or exception o Replaced the word "process" with "thread" in comments. "process" implies some measure of isolation (namespace, heap, file descriptors, etc.) whereas "thread" just impliues a separate thread of control. These are _not_ processes! [Hey, I work on operating systems these days - they're fussy about these definitions.] Revision Changes Path 1.8 +8 -3 hugs98/lib/Prelude.hs 1.4 +10 -10 hugs98/lib/exts/ConcBase.hs

1 0

cvs commit: hugs98/lib/exts Exception.hs
by Alastair Reid 03 Apr '01

03 Apr '01

reid 2001/04/03 12:19:10 PDT Modified files: lib/exts Exception.hs Log: Added Exception.finally Revision Changes Path 1.2 +4 -0 hugs98/lib/exts/Exception.hs

1 0

cvs commit: hugs98/src type.c
by Alastair Reid 02 Apr '01

02 Apr '01

reid 2001/04/02 20:00:07 PDT Modified files: src type.c Log: Applied Julian Seward's bugfix (from Feb 25th) Revision Changes Path 1.31 +3 -2 hugs98/src/type.c

1 0

cvs commit: hugs98/src parser.y
by Alastair Reid 02 Apr '01

02 Apr '01

reid 2001/04/02 19:56:36 PDT Modified files: src parser.y Log: Applied Sigbjorn's bugfix for the parser. (Without this fix, you can't load lib/exts/Int.hs. I guess it should be added to the mythical test suite. Which reminds me... what happened to the test suite? We used to have one. I think it was in hugs/src/tests. A (admittedly superficial) search failed to reveal it.) Revision Changes Path 1.16 +11 -6 hugs98/src/parser.y

1 0

RE: cvs commit: hugs98/lib Prelude.hs hugs98/lib/exts Exception.hs ConcBase.hs
by Simon Marlow 02 Apr '01

02 Apr '01

> One big difference is that, because IOError handling is handled > separately from HugsExceptions, we can't readily implement > > throw :: Exception -> a > > But we can provide > > throwIO :: Exception -> IO a > > (which I'll try to persuade the GHC people to ask to their > implementation). This is called 'ioError' in GHC, because the Exception type is now a synonym for IOError (in GHC 5.00). This makes it easier to port code that uses IOErrors to use Exceptions - the Exception interface is just an extension of the existing IOError interface. Cheers, Simon

1 0

RE: cvs commit: hugs98/src builtin.c
by Simon Marlow 02 Apr '01

02 Apr '01

> reid 2001/04/01 21:24:13 PDT > > Modified files: > src builtin.c > Log: > This is a bogus commit because I just accidentally > committed all files > instead of just machdep.c (and I can't remember the cvs > magic to undo > it). $ cvs up -j<new> -j<old> <file1> $ cvs up -j<new> -j<old> <file2> $ etc. $ cvs commit -m "undo bogus commit" <file1> <file2> ... where <new> and <old> are the new and old revisions of the file (eg. if the last commit was revision 1.3, then <new>=1.3 and <old>=1.2). Alternatively, use a "forced commit" (cvs commit -f) to add the correct log message to each of the committed files. Cheers, Simon

1 0

cvs commit: hugs98/lib Prelude.hs hugs98/lib/exts Exception.hs ConcBase.hs
by Alastair Reid 02 Apr '01

02 Apr '01

reid 2001/04/01 21:51:23 PDT Modified files: lib Prelude.hs lib/exts ConcBase.hs Added files: lib/exts Exception.hs Log: Even if the previous commits are bad, things should still work because I haven't changed anything in the normal evaluation path. This set of commits had better work though! Prelude.hs: There's now two kinds of exception: IOError and HugsException. IOErrors are propagated using the normal continuation passing mechanisms - exception handlers are chained together on the heap. (This stuff is unchanged. Changing that would require changes to greencard and every IO primop in Hugs.) HugsExceptions are propagated using longjmp, etc. Exception handlers are piled up on the stack. But, what happens when we have to do a context switch (e.g., a call to Concurrency.yield or Concurrency.takeMVar)? We have to remember all the HugsException catches so that we can restore them when we switch back to that thread. This is what Hugs_Catch is for - it remembers which exception handlers we have to reinstall on the stack when we switch back to the thread. (btw, The type names, constructors, etc. associated with IOResult, etc. really ought to be renamed so they have the word "Hugs" or "prim" in them for the usual namespace-contamination reasons.) ConcBase.hs: Some minor organisation and layout changes - trying to restore a consistent style. Also, forkIO has to make sure that every thread contains both an IOError handler and HugsException handler. new file exts/Exceptions.hs: A (very) cut-down implementation of GHC's Exception module. One big difference is that, because IOError handling is handled separately from HugsExceptions, we can't readily implement throw :: Exception -> a But we can provide throwIO :: Exception -> IO a (which I'll try to persuade the GHC people to ask to their implementation). The other big difference is that GHC's Exception datatype has many constructors for different kinds of errors. Implementing this would require extensive changes to any primops which can raise exceptions - lots of changes. So, instead, you get a HugsException and you can convert it to a string and, if you see fit, try to parse the string. I didn't add GHC's assertions - so they're left out. Adding Dynamic exceptions looked tricky and didn't seem like a pressing need so I left them out. Revision Changes Path 1.7 +32 -1 hugs98/lib/Prelude.hs 1.3 +14 -9 hugs98/lib/exts/ConcBase.hs

1 0

cvs commit: hugs98/src builtin.c
by Alastair Reid 02 Apr '01

02 Apr '01

reid 2001/04/01 21:24:13 PDT Modified files: src builtin.c Log: This is a bogus commit because I just accidentally committed all files instead of just machdep.c (and I can't remember the cvs magic to undo it). So here's what all those committed files did: interns.c: The old file interns.c (which is usually ifdef'd out) provides access to a bunch of Hugs internals. This provided a primitive: catchException2 :: a -> Either Cell a where Cell is an error object. The only change here is a fix to the implementation. It couldn't possibly work the way it was - which leaves me wondering what happened to it. builtin.c: Put a (fixed) copy of catchException2 here and added throwException data HugsException primitive primCatchException :: a -> Either HugsException a primitive primThrowException :: HugsException -> a Note that using these directly would break referential transparency. To fix that, one or the other (doesn't matter) ought to be in the IO monad. But there's difficult interactions with the thread implementation (that's what the fuss is about after all!) so we provide the unsafe versions here and wrap them up before exporting them. Incidentally, notice that these names (which have to be exported from the Prelude and so pollute the user's namespace) either include the word "hugs" or starts with the word "prim". printer.c: Added a primop for printing Hugs exception objects primitive primShowException :: HugsException -> String It originally had type: primitive primShowException :: HugsException -> ShowS but printDBadRedex ignores its second argument when printing error objects. Revision Changes Path 1.12 +3 -3 hugs98/src/builtin.c

1 0

cvs commit: hugs98/src builtin.c interns.c machdep.c printer.c storage.h
by Alastair Reid 02 Apr '01

02 Apr '01

reid 2001/04/01 21:05:13 PDT Modified files: src builtin.c interns.c machdep.c printer.c storage.h Log: Minor cleanup to kill a gcc warning. If you use /* ... */ to comment out code that includes a comment, gcc will complain. If you use #if 0 ... #endif, to comment out code, it nests properly and gcc has no cause for complaint. Revision Changes Path 1.11 +43 -5 hugs98/src/builtin.c 1.5 +3 -3 hugs98/src/interns.c 1.15 +6 -4 hugs98/src/machdep.c 1.5 +16 -4 hugs98/src/printer.c 1.18 +2 -4 hugs98/src/storage.h

1 0