Граф коммитов

1081 Коммитов

Автор SHA1 Сообщение Дата
brendan%mozilla.org 20e8384705 Improve a comment that implied a need to include jsstr.h. 2000-09-05 21:39:48 +00:00
brendan%mozilla.org 53ea2141f3 Must JS_SetGlobalObject at least, if not JS_InitStandardClasses, before defining native functions at startup. 2000-09-04 00:10:06 +00:00
scc%mozilla.org 37f44817be fixing bustage exposed by removing extra comparisons 2000-09-03 03:28:05 +00:00
brendan%mozilla.org 57422a219a Support external string GC types so XPConnect and other systems can optimize away copies (50602, r=jband). 2000-09-02 20:37:37 +00:00
jband%netscape.com a1659d2bd2 fix crasher bug 51176. xpti semantics for GetParent changed sometime back and I did not find all the cases where this mattered. thanks to mang@subcarrier.org. a,r=brendan@mozilla.org 2000-09-02 20:30:56 +00:00
brendan%mozilla.org f4f8bbb650 Followup for 44997, r=shaver:
- #if JS_HAS_LVALUE_RETURN around cx->rval2/rval2set defs and uses.
- Instrument different kinds of invocations, #ifdef DEBUG only.
- Clean up basis case of empty switch statement to use high = -1, low = 0,
  requiring care when optimizing in-range tests using unsigned casts, but
  freeing the interpreter and decompiler from having to do an extra test
  before looping from low to high.
- Clean up all codegen to use JUMP_OFFSET_LEN, ATOM_INDEX_LEN, etc. instead of
  magic 2 or 4.
- Add JSOP_TRY and JSOP_FINALLY no-ops to save a srcnote per JSOP_NOP, and to
  make decompilation and jit'ing easier.
- Minimize number of source notes to maximize SRC_XDELTA span.
- Use JSSCRIPT_FIND_CATCH_START in throw code.
- Indentation and bracing nits picked.
2000-09-01 18:42:22 +00:00
brendan%mozilla.org e185d5b9b7 Fix js_EnterSharpObject to clean up sharpObjectMap on error (44009, r=shaver). 2000-09-01 18:01:04 +00:00
jband%netscape.com a7289b6668 fix bug 50941 by updating help text. r=shaver@mozilla.org a=brendan@mozilla.org 2000-09-01 01:02:39 +00:00
brendan%mozilla.org 38e38f2196 Support ECMA reference type return from native methods (44997, r=shaver). 2000-09-01 00:45:50 +00:00
waldemar%netscape.com 191f479b10 Added jsdhash.c 2000-08-31 20:41:33 +00:00
jband%netscape.com 8b20a44782 fix to un-bust XPCONNECT_STANDALONE build (#ifdef'd out code for the browser) from shaver@mozilla.org 2000-08-31 10:44:28 +00:00
brendan%mozilla.org a951a13f16 Fix GC_MARK_DEBUG bustage (r=dbaron, not part of build unless you -D it). 2000-08-31 03:41:36 +00:00
brendan%mozilla.org 2b858a9a05 Forgot to round capacity up to nearest power of two\! r=bienvenu. 2000-08-31 03:12:40 +00:00
beard%netscape.com ed0176a047 bug #44187: added guards to prevent crashes when Java not available. r=jband 2000-08-31 02:37:42 +00:00
brendan%mozilla.org b4a84c6ca7 Script to NSPR-ize and 'PL-ize' jsdhash.[ch] so they fit (sort of) over in xpcom/ds. 2000-08-30 19:30:33 +00:00
mkaply%us.ibm.com a7e2c5bb4c #50212
r=brendan, a=brendan
OS/2 GCC doesn't define _M_IX86
2000-08-30 01:04:59 +00:00
brendan%mozilla.org 20a51b8298 Finish fixing 42221, for sure. 2000-08-29 23:09:47 +00:00
mccabe%netscape.com 5c644e8fc3 Potential fix to 50313
Initialize stack variable to NULL on suspicion that later possibly-unset use of it is causing crash.
2000-08-26 07:10:31 +00:00
brendan%mozilla.org c77f05ae2e Clean up JSObjectOps layering violations by adding mark and clear ops; JSClass gets a corresponding mark op so classes with unregistered roots in private data can mark them. The JS API gets a new JS_MarkGCThing entry point for JSObjectOps.mark implementors. Prerequisite check-in for bug 49816 and others (r=shaver). 2000-08-26 02:30:22 +00:00
brendan%mozilla.org 9f6b515501 Lazily define escape and unescape (via js_InitStringClass) only #ifndef MOZILLA_CLIENT, because the DOM must predefine backward-compatible versions of these functions that 'stick' (42221, r=mccabe). 2000-08-26 02:05:41 +00:00
mccabe%netscape.com 91dae5c247 Fix to 39141.
Check for empty element case in array literals ( first element in [,'foo'] ) now uses the next token instead of the previous one when constructing the node, so the first element gets TOK_COMMA instead of TOK_LB.

This fixes a crash from previously accepted JS.

r=rogerl
2000-08-26 01:44:35 +00:00
brendan%mozilla.org 53dfa73056 Macro-ize FPU mode fixing and do it on ever js_InitNumberClass; also, simplify #ifdefs to unify OS2 with XP_PC && _M_IX86 case (50212, r=mkaply@us.ibm.com). 2000-08-25 03:14:50 +00:00
jband%netscape.com 47af0b5999 bug 49748. Fix crash due to JS loader holding xpconnect service too long. Also add unimplemented method to xpconnect service and fix cast for bug 49641. r=shaver@mozilla.org a=brendan@mozilla.org 2000-08-24 07:50:27 +00:00
beard%netscape.com 85ec8a2278 use QueryElementAt() to save space and time. r=shaver 2000-08-24 04:10:47 +00:00
brendan%mozilla.org f80865ab42 Eliminate silly gcc warning. 2000-08-24 03:14:17 +00:00
brendan%mozilla.org 0b6d9aa649 Add strict warning about trailing comma in object initializers being non-ECMA (50001, r=shaver). 2000-08-23 23:55:43 +00:00
rogerl%netscape.com d876d27a02 r,a=beard@netscape.com. Bug #45415, fix LiveConnect memory leak. 2000-08-22 23:31:09 +00:00
brendan%mozilla.org f447340dff Fix UMR on ok in JS_ResolveStandardClass (46703, r=rogerl). 2000-08-22 00:25:17 +00:00
pavel%gingerall.cz d24db69aaf - JSVALToSV handles perl objects in the right way 2000-08-21 16:52:49 +00:00
brendan%mozilla.org 6aea0bf81f Re-constipate JSClass.name. 2000-08-19 19:17:32 +00:00
jband%netscape.com 577d93dfb4 fix the warnings that went along with the last bustage fix - in case one of the compilers needs it. brendan can fix as he likes in the morning 2000-08-19 10:36:45 +00:00
jband%netscape.com 598c2a90c2 fix bustage 2000-08-19 10:22:13 +00:00
brendan%mozilla.org 56c94e849a Move shell to DIRS from ENABLE_TESTS ifdef addition to DIRS (r=mccabe,leaf). 2000-08-19 08:41:34 +00:00
brendan%mozilla.org f59c35edb2 Support lazy initialization of standard classes and their associated global functions/properties (46703, r=jband,rogerl). 2000-08-19 08:37:07 +00:00
brendan%mozilla.org 896fb216bc Add stub entry struct and ops for const void *keys, split out JS_DHashTableRawRemove, beefed up comments (part of 46703, r=jband). 2000-08-19 08:36:32 +00:00
cls%seawood.org 5ef043ad42 Use the -n32 ABI as the default for Irix 6.x but allow the option to disable it. 2000-08-19 08:05:05 +00:00
jband%netscape.com 77fff625c9 Add support for -w and -s for reporting warnings and doing strict JS checking (as supported by the engine). Also update the usage info. ra=brendan@mozilla.org 2000-08-19 03:28:01 +00:00
jband%netscape.com 870adce9bf Adding better diagnostic message to help in finding and fixing mismatched calls to JS_AddNamedRoot/JS_RemoveRoot. ra=brendan@mozilla.org 2000-08-19 03:25:16 +00:00
brendan%mozilla.org 95c4e4dc86 Add jsdhash. 2000-08-18 00:04:53 +00:00
rogerl%netscape.com 2799c5b4ed Backing out fix to see if Linux build passes alivetest. 2000-08-17 00:30:57 +00:00
rogerl%netscape.com d3804fb66f Bug #31255. r,a=brendan. Once more unto the '-->' accepted as line comment
fix.
2000-08-16 22:46:04 +00:00
rogerl%netscape.com e1584cd317 Bug #39141, r=mccabe, a=beard. Fix crash for empty array initializer. 2000-08-16 22:43:26 +00:00
shaver%mozilla.org f66cde2438 Fix 47354 and 39975 by providing a system-privileged scope backstop for
JS Components, and teaching the ScriptSecurityManager to check for
XPC-wrapped native objects in the scope chain when looking for an
object's principal. r=jband/a=brendan
2000-08-16 04:01:02 +00:00
mccabe%netscape.com d8cabef55b Fix to 39438.
Add a TokenStream entry to the regexp parser state so that regexps compiled at script-compilation time can report filename and line number on error.
2000-08-11 23:51:12 +00:00
conrad%ingress.com d9e8edcf29 Changed from using strings for nsIDirectoryService props to using macro names defined in nsDirectoryServiceDefs.h r=valeski 2000-08-11 20:31:57 +00:00
mccabe%netscape.com 16ab7000d2 Back in backed-out fix. 2000-08-11 02:58:30 +00:00
mccabe%netscape.com e38da88c9f Backout until I can figure out why mac doesn't like strlen here, even though it seems OK elsewhere. 2000-08-10 23:32:03 +00:00
mccabe%netscape.com 8ddfc050c7 Fix to 46566.
Always copy the current line string out of the token buffer when generating an error report, rather than just passing the token buffer itself.  The token buffer wasn't necessarily a well-terminated string, so displaying the contents of the string in the error report produced unexpected results.

The unicode string in the error report is owned by a JSString; this string is rooted for the (stack-based) lifetime of the error report.

Fix courtesy jband.

r=mccabe
a=beard
2000-08-10 23:02:39 +00:00
mccabe%netscape.com bb5465f725 Fix to 34425 -
Always create an error report, even when there is no current stack frame and it might be empty.  This fixes an API regression; we used to allow JS_ReportError to be called from the API when no JavaScript was running.

r=rogerl
a=beard
2000-08-10 22:57:40 +00:00
mccabe%netscape.com 69b8cc3cca Fix to 47409 -
Don't use timezone comment in javascript Date object toString string if it looks like it might not be ASCII.

a,r=beard
2000-08-10 22:54:45 +00:00