summaryrefslogtreecommitdiffstats
path: root/generic
Commit message (Collapse)AuthorAgeFilesLines
* Convert the "bytearray" Tcl_ObjType to use the proposed Tcl_ObjIntReptip_445_rejectdgp2016-03-261-46/+63
| | | | | manipulation routines. This works, but requires too much ugliness. The ugliness reveals that the interface needs some refinement, making this a rejected branch of development.
* irPtr = NULL passed to Tcl_StoreIntRep clears out any value for typePtr.dgp2016-03-251-4/+7
|
* Get signatures in sync.dgp2016-03-251-1/+1
|
* First demonstration conversion to the new intrep manipulation routines.dgp2016-03-253-30/+37
|
* New routines Tcl_FetchIntRep() and Tcl_StoreIntRep().dgp2016-03-254-0/+94
|
* Create a type Tcl_ObjIntRep so we can pass intreps as arguments.dgp2016-03-251-20/+24
|
* merge trunkdgp2016-03-241-80/+3
|\
| * No longer a need to keep around 'revert to 8.5' code.dgp2016-03-241-80/+3
| |
* | Revise the "index" Tcl_ObjType to use the new routine.dgp2016-03-241-7/+1
| |
* | Revise the "instname" Tcl_ObjType to use the routines.dgp2016-03-241-9/+10
| |
* | oopsdgp2016-03-241-1/+1
| |
* | Update Tcl_InitStringRep callers to handle OOM condition.dgp2016-03-246-5/+28
| |
* | Revise Tcl_InitStringRep() to do non-panic attempt at allocation.dgp2016-03-242-18/+10
| | | | | | | | Let caller decide how catastrophic it is. Revise [string repeat] to use new routine.
* | More purging of direct access to bytes field where it isn't important.dgp2016-03-241-14/+4
| |
* | Update more Tcl_ObjTypes to use Tcl_InitStringRep().dgp2016-03-242-35/+26
| | | | | | | | Adapt TclInitStringRep macro to accept TclInitStringRep(objptr, NULL, 0) without warning -- requires outwitting compiler.
* | TclInitStringRep() already knows about tclEmptyStringRep.dgp2016-03-242-2/+2
| |
* | oopsdgp2016-03-242-5/+3
| |
* | stay out of internals when nice interfaces are available.dgp2016-03-241-8/+2
| |
* | Revised "end-offset" Tcl_ObjType to use new routine.dgp2016-03-241-8/+7
| |
* | Revise "ensembleCommand" Tcl_ObjType to use new routines.dgp2016-03-231-4/+2
| |
* | Convert "dict" Tcl_ObjType to use new routines.dgp2016-03-231-12/+15
| |
* | Tcl_InitStringRep() bug. Truncation assumed length == allocated. Wrong!dgp2016-03-232-37/+31
| | | | | | | | Convert "bytearray" Tcl_ObjType to used new facilities. No longer directly refers to bytes or length fields, or any ckalloc of string rep.
* | Revise Tcl_InitStringRep(); numBytes is unsigned. Only truncation permitted.dgp2016-03-233-14/+16
| |
* | Release memory after truncation.dgp2016-03-231-0/+5
| |
* | Make sure no path emerges to write on tclEmptyStringRep.dgp2016-03-231-0/+1
| |
* | Next step: new routine Tcl_InitStringRep()dgp2016-03-234-1/+90
| |
* | First simple step implementing TIP 445.dgp2016-03-224-0/+38
|/
* Factor out string internal rep definition so fix for [1af8de570511] is less ↵dkf2016-03-223-58/+102
|\ | | | | | | awful.
| * Factor out string internal rep definition so fix for [1af8de570511] is less ↵dkf2016-03-223-58/+102
| | | | | | | | awful.
* | Revised fix for signed/unsigned mismatch. "get types right" > "add more casts"dgp2016-03-224-9/+9
|\ \ | |/
| * More places where unsigned quantities should be decared such.dgp2016-03-222-2/+2
| |
| * More signed/unsigned type correction.dgp2016-03-221-3/+3
| |\
| | * (cherry-pick): Fix signed-unsigned-compare warning (reported by François ↵jan.nijtmans2016-03-222-2/+2
| | | | | | | | | | | | Vogel on Windows, but gcc can trigger it as well)
| * | Don't compare signed and unsigned. Get types right.dgp2016-03-221-2/+2
| |/
* | Fix signed-unsigned-compare warning (reported by François Vogel on Windows, ↵jan.nijtmans2016-03-222-2/+2
| | | | | | | | but gcc can trigger it as well)
* | [1af8de570511] Fix crash in [string replace] caused by cut-n-paste.dkf2016-03-201-22/+13
|\ \ | |/
| * [1af8de570511] Fix crash in [string replace] caused by cut-n-paste.dkf2016-03-201-22/+13
| |
* | [0b8c387cf7] Replace deprecated Tcl_VarEval(). socket accept callbacks nowdgp2016-03-171-20/+25
|\ \ | | | | | | always evaluate in global context. *** POTENTIAL INCOMPATIBILITY ***
| * \ Merge trunk. bug_0b8c387cf7jan.nijtmans2016-03-165-58/+39
| |\ \ | | | | | | | | Add TCL_EVAL_GLOBAL flag to Tcl_EvalObjEx(), for reason mentioned in [0b8c387cf7].
| * | | [0b8c387cf7] Replace deprecated Tcl_VarEval() call with reworked callbackdgp2016-03-081-20/+25
| | | | | | | | | | | | system that uses Tcl_Obj scripts.
* | | | merge 8.6 bug fixesdgp2016-03-173-12/+56
|\ \ \ \ | |_|/ / |/| | / | | |/ | |/|
| * | [ae38befcfb] Rewrite TclGetInnermostExceptionRange() for fewer iterations.dgp2016-03-171-11/+10
| | |
| * | [1a25fdfec6] Enable compile of commands like [variable ${ns}::v].dgp2016-03-171-1/+3
| |\ \
| | * | Simple change gets most of the effect. Fails to handle backslash. anyone care?bug_1a25fdfec6dgp2016-03-111-1/+3
| | | |
| * | | [b9b2079e6d] Proposed fix. When a compileProc fails, it may have done andgp2016-03-101-0/+43
| |/ / | | | | | | | | | | | | arbitrary amount of partial work, which needs to be undone. When the exception handling machinery got its last big revision, the undoing of what it does was neglected. I think this patch gets it all, but more eyes would be good.
* | | Remove undocumented TCL_HASH_KEY_STORE_HASH. Setting this to "0" is not ↵jan.nijtmans2016-03-162-55/+0
| | | | | | | | | | | | supported (was it ever ????), and not tested for long time.
* | | Allow unstable Tcl releases to use unstable packages by default (this makes ↵jan.nijtmans2016-03-112-2/+38
|\ \ \ | |_|/ |/| | | | | "package require Tk" find the expected 8.7a0 version in stead of 8.6). Make many "package" test-cases work even when the environment variable TCL_PKG_PREFER_LATEST is set.
| * | Let Tcl 8.7 allow Tk 8.7 to be used by defaultjn_Tcl_requirementjan.nijtmans2016-03-091-1/+1
|/ /
* | [bbc304f61a] Avoid event handling when reflected channel has a watch change ↵dgp2016-03-081-2/+2
|\ \ | |/ | | | | half-completed. (First half in 1 thread, second in another). When this is allowed to happen, false alarm errors from [chan postevent] are the result when timing is unlucky. See test iocmd.tf-24.17.bug3522560 .
| * [bbc304f61a] Avoid event handling when reflected channel has a watchdgp2016-03-081-2/+2
| |\ | | | | | | | | | | | | change half-completed. (First half in 1 thread, second in another). When this is allowed to happen, false alarm errors from [chan postevent] are the result when timing is unlucky. See test iocmd.tf-24.17.bug3522560 .
generated by cgit v0.12 at 2025-04-22 08:41:23 (GMT)