Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Merge 8.5 | jan.nijtmans | 2019-09-10 | 14 | -98/+98 |
|\ | |||||
| * | Don't use constraints like unixOrPc anymore, use unixOrWin (for example) | jan.nijtmans | 2019-09-10 | 12 | -133/+142 |
| | | |||||
| * | Backport some improvements to tm.tcl (mostly comments). | jan.nijtmans | 2019-09-10 | 4 | -126/+112 |
| | | | | | | | | Don't use ::tcl_platform(debug) anymore, since it cannot be thrusted: Better use [::tcl::pkgconfig get debug] Reduce limits in tests/compile.test (13.2), since apparently it's still too much for some platforms. | ||||
* | | Merge 8.5 | jan.nijtmans | 2019-09-09 | 1 | -6/+6 |
|\ \ | |/ | |||||
| * | Move &crosstest up in .travis.yml | jan.nijtmans | 2019-09-09 | 1 | -6/+6 |
| | | |||||
* | | Merge 8.5 | jan.nijtmans | 2019-09-09 | 2 | -1/+7 |
|\ \ | |/ | |||||
| * | Final touch to make everything work for 8.5: | jan.nijtmans | 2019-09-09 | 2 | -4/+4 |
| | | | | | | | | | | - Don't use AS_ECHO macro, because autoconf-2.59 doesn't have it. - -D_BUILD_tcl should be -DBUILD_tcl - Missing back-slashes at the end of the line. | ||||
* | | Merge 8.5 | jan.nijtmans | 2019-09-08 | 1 | -36/+36 |
|\ \ | |/ | |||||
| * | dde and registry extension should be compiled with -DUNICODE -D_UNICODE. | jan.nijtmans | 2019-09-08 | 2 | -37/+45 |
| | | | | | | Put 64-bit builds before 32-bit builds in travis | ||||
* | | Merge 8.5 | jan.nijtmans | 2019-09-07 | 1 | -1/+1 |
|\ \ | |/ | |||||
| * | Add --enable-threads to Windows/GCC (native) builds. This is - most likely - ↵ | jan.nijtmans | 2019-09-07 | 1 | -3/+3 |
| | | | | | | | | the cause of the travis failure in compile.test. | ||||
* | | Don't let Tcl compilation depend on USE_32BIT_TIME_T any more: Microsoft ↵ | jan.nijtmans | 2019-09-07 | 4 | -22/+25 |
|\ \ | |/ | | | | | | | could discontinue this macro any moment, then we are prepared .... As a bonus: time_t is now allowed to be 64-bit internally, without effect on the C API (like stub-enabled extensions) | ||||
| * | Don't let Tcl compilation depend on USE_32BIT_TIME_T any more: Microsoft ↵ | jan.nijtmans | 2019-09-07 | 5 | -23/+26 |
| |\ | | | | | | | | | | | | | could discontinue this macro any moment, then we are prepared .... As a bonus: time_t is now allowed to be 64-bit internally, without effect on the C API (like stub-enabled extensions) | ||||
| | * | Don't let Tcl depend on USE_32BIT_TIME_T any more: If your compiler supports ↵ | jan.nijtmans | 2019-09-06 | 4 | -22/+25 |
| | | | | | | | | | | | | | | | it, time_t will be 64-bit internally. But at API-level, time_t will still be restricted to 32-bit on Win32 (Not on Win64). This keeps Tcl_StatBuf the same (unless USE_64BIT_TIME_T is defined), so 64-bit times still cannot be used everywhere. | ||||
* | | | Fix bug 9d10c37aa8 (in the Tk repository): Improperly converted link in HTML ↵ | fvogel | 2019-09-07 | 1 | -2/+2 |
| | | | | | | | | | | | | man page for ttk::style | ||||
* | | | merge 8.5 | sebres | 2019-09-06 | 2 | -3/+8 |
|\ \ \ | |/ / | |||||
| * | | if frameName (actual level) does not contain a real level (#0 or 1) ↵ | sebres | 2019-09-06 | 2 | -4/+9 |
| | | | | | | | | | | | | | | | historically TclGetFrame and Tcl_UpVar2 uses current level - 1, so to put supplied name in case of bad level (error at top - 1) is wrong; be more consistent with TclObjGetFrame (at least in error case if relative level used). | ||||
* | | | merge 8.5 (no functional changes on tclProc, it was already fixed there - ↵ | sebres | 2019-09-06 | 4 | -10/+61 |
|\ \ \ | |/ / | | | | | | | only test cases) | ||||
| * | | fix bad level (if specified argument is not a level at all) | sebres | 2019-09-06 | 1 | -0/+1 |
| | | | |||||
| * | | cherry-pick [b87d2183ca]: test cases covering bug [775ee88560]: segfault in ↵ | sebres | 2019-09-06 | 2 | -0/+21 |
| | | | | | | | | | | | | upvar at wrong level, wrong message of uplevel | ||||
| * | | Fix configure script (re-generated with a modified autoconf-2.59, in which ↵ | jan.nijtmans | 2019-09-06 | 2 | -38/+70 |
| |/ | | | | | | | | | the AC_PROG_MAKE_SET macro is replaced with the one from autoconf-2.69) :-) Re-enable native travis build on Windows, showing that the build now works. | ||||
* | | Simplify searching for tcl86.lib (and related files): First search for the ↵ | jan.nijtmans | 2019-09-06 | 1 | -15/+9 |
| | | | | | | | | one without 't' suffix, then the 't' variant. (Without 't' = built with 'configure'/'make', with 't' = built with 'nmake') | ||||
* | | Fix [579a05fb34] (partly): b) tcltest file has mismatched version number. | jan.nijtmans | 2019-09-06 | 2 | -4/+4 |
| | | |||||
* | | merge 8.5 | sebres | 2019-09-05 | 2 | -2/+7 |
|\ \ | |/ | |||||
| * | amend to [4718b41c56]: check size of st_mtime instead of time_t in constraint | sebres | 2019-09-05 | 2 | -2/+7 |
| | | |||||
* | | Fix [4718b41c56]: windows x86 & x64: file mtime overflows in modification ↵ | jan.nijtmans | 2019-09-05 | 1 | -1/+1 |
| | | | | | | | | date (2038?, windows 32-bit time_t?) | ||||
* | | Don't build on travis with GCC on Windows, since it currently doesn't work ↵ | jan.nijtmans | 2019-09-04 | 1 | -29/+2 |
| | | | | | | | | in combination with autoconf-2.59-generated configure | ||||
* | | When using Tcl 8.7 headers, don't worry about threaded-allocator mismatch ↵ | jan.nijtmans | 2019-09-04 | 1 | -3/+3 |
| | | | | | | | | any more in rules.vc. | ||||
* | | Docfix: \0 is special in nroff, so use \e0 instead | dkf | 2019-09-03 | 1 | -3/+3 |
| | | |||||
* | | Backout [87d28ec30a29e3ce]: Looks like it causes test-failures in event.test ↵ | jan.nijtmans | 2019-09-03 | 1 | -2/+1 |
|\ \ | |/ | | | | | on Windows. | ||||
| * | Backout last commit: Looks like it causes test-failures in event.test on ↵ | jan.nijtmans | 2019-09-03 | 1 | -2/+1 |
| | | | | | | | | Windows. | ||||
* | | Fix testing for debug build on UNIX too (on UNIX, ::tcl_platform(debug) is ↵ | jan.nijtmans | 2019-09-02 | 1 | -1/+1 |
| | | | | | | | | not set ....) | ||||
* | | windows (mingw build): fix debug recognition (::tcl_platform(debug)), no ↵ | sebres | 2019-09-02 | 2 | -2/+2 |
| | | | | | | | | debug if NDEBUG is set | ||||
* | | Merge 8.5 | jan.nijtmans | 2019-09-02 | 1 | -1/+2 |
|\ \ | |/ | |||||
| * | Missing TCL_GLOBAL_ONLY flag in VwaitVarProc(): vwait always references ↵ | jan.nijtmans | 2019-09-02 | 1 | -1/+2 |
| | | | | | | | | global variables, this could lead to strange side-effects. | ||||
* | | Merge 8.5. Add (still unused) runstatedir in Makefile.in. Re-generate ↵ | jan.nijtmans | 2019-09-02 | 4 | -3290/+3124 |
|\ \ | |/ | | | | | win/unix configure scripts. | ||||
| * | Fix [5591e4a820]: @TCL_EXE@ not properly generated in 8.5 branch. Added ↵ | jan.nijtmans | 2019-09-02 | 2 | -2/+3 |
| | | | | | | | | @runstatedir@ in Makefile.in, not used yet (except if someone decides to re-generate "configure" with autconf-2.70). | ||||
* | | merge 8.5 - compile.test: reduce count of nested scripts to 1000 in debug ↵ | sebres | 2019-08-30 | 1 | -2/+2 |
|\ \ | |/ | | | | | | | case (seems to be to heavy on some platforms within debug-build); (small amend to the fix of [fec0c17d39]) | ||||
| * | compile.test: reduce count of nested scripts to 1000 in debug case (seems to ↵ | sebres | 2019-08-30 | 1 | -2/+2 |
| | | | | | | | | | | be to heavy on some platforms within debug-build); (small amend to the fix of [fec0c17d39]) | ||||
* | | fixes bug [fec0c17d39] (regression firstly introduced in [bf171b6b51]): ↵ | sebres | 2019-08-29 | 2 | -18/+60 |
|\ \ | | | | | | | | | | stack overflow (followed by segfault) by compilation of too many nested scripts (don't use system stack anymore) | ||||
| * | | closes [fec0c17d39]: fixed stack overflow (followed by SF) by compilation of ↵ | sebres | 2019-08-29 | 1 | -18/+29 |
| | | | | | | | | | | | | too many nested bodies (don't use system stack, size of Tcl_Parse is ca. 400 bytes and compiler proc's of commands are reserving stack too) | ||||
| * | | more variants in test (since 8.6 compiles "try" using evalStk instruction) | sebres | 2019-08-29 | 1 | -2/+2 |
| | | | |||||
| * | | bug [fec0c17d39]: add test cases covering nested compilation (8.6th compile ↵ | sebres | 2019-08-29 | 1 | -0/+31 |
| |\ \ | | |/ | | | | | | | generates a stack overflow with following segfault by 2000 nested bodies) | ||||
| | * | add test cases covering nested compilation bug [fec0c17d39] (8.5 is not ↵ | sebres | 2019-08-29 | 1 | -0/+31 |
| | | | | | | | | | | | | affected at the moment by nested count under 2500) | ||||
* | | | Bug fix 889065786b. Add stubs related flags when compiling extension stubs. | apnadkarni | 2019-08-29 | 1 | -3/+3 |
| | | | |||||
* | | | Starting with Tcl 8.7, TCL_THREADS is 1 by default. Adapt rules.vc for that. | jan.nijtmans | 2019-08-29 | 1 | -5/+5 |
| | | | |||||
* | | | Add /* FALLTHRU */ markers in various places (silencing possible GCC ↵ | jan.nijtmans | 2019-08-28 | 23 | -61/+67 |
|/ / | | | | | | | warnings). Eliminate some more "register" keywords. Eliminate (or silence) some unused function parameters. | ||||
* | | closes [fa6bf38d07]: integrate bug-fa6bf38d07-v2 | sebres | 2019-08-27 | 3 | -16/+116 |
|\ \ | |||||
| * | | small amend: be sure checkInterp is set if entering back the code marked as ↵ | sebres | 2019-08-27 | 1 | -2/+2 |
| | | | | | | | | | | | | TCL_BYTECODE_RECOMPILE (normally also set in CACHE_STACK_INFO, but...) | ||||
| * | | simplification, use the same "fixed" (and faster) code for INST_EVAL_STK | sebres | 2019-08-27 | 1 | -11/+6 |
| | | |