Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Preparation to provide TclNRPackageObjectCmd: Eliminate the loop in | pooryorick | 2018-02-12 | 2 | -262/+264 |
| | | | PkgRequireCore so that TclNRAddCallback can be added at the needed spots. | ||||
* | Change the signature of PkgRequireCore in preparation to provide | pooryorick | 2018-02-11 | 1 | -22/+21 |
| | | | TclNRPackageObjCmd. | ||||
* | Remove restriction on defining the class of a TclOO object not explicitly | pooryorick | 2018-02-11 | 1 | -15/+0 |
| | | | instantiated from ::oo::class. | ||||
* | typo's (found by Gustaf Neumann). Thanks! | jan.nijtmans | 2018-01-17 | 2 | -8/+8 |
| | |||||
* | Fix behavior of Tcl_GetRange() and "string range" regarding surrogates, when ↵ | jan.nijtmans | 2018-01-11 | 1 | -1/+12 |
| | | | | Tcl is compiled with -DTCL_UTF_MAX=4. Partial fix for bug [11ae2be95dac9417]. Also, fix typo. | ||||
* | Fix bug introduced in [0dd0d14489258621] (only for TCL_UTF_MAX > 3): If len ↵ | jan.nijtmans | 2017-12-28 | 1 | -0/+2 |
| | | | | | parameter = -1, returned size includes terminating 0-byte. So, account for that. Also, fix some comments which were not accurate any more, now that Windows 95/98 is not supported any more. | ||||
* | Fix handling of surrogates (when TCL_UTF_MAX > 3) in ↵ | jan.nijtmans | 2017-12-28 | 2 | -31/+32 |
| | | | | Tcl_UtfNcmp()/Tcl_UtfNcasecmp()/TclUtfCasecmp(). Backported from core-8-branch, where this was fixed already. | ||||
* | Make Tcl_WinTCharToUtf/Tcl_WinUtfToTChar work when TCL_UTF_MAX > 3, mainly ↵ | jan.nijtmans | 2017-12-28 | 1 | -1/+33 |
| | | | | when sizeof(Tcl_UniChar) != 2. | ||||
* | merge release | jan.nijtmans | 2017-12-23 | 1 | -2/+2 |
|\ | |||||
| * | merge 8.6; update changes | dgp | 2017-12-19 | 1 | -1/+1 |
| |\ | |||||
| * \ | merge 8.6 | dgp | 2017-12-06 | 1 | -0/+4 |
| |\ \ | |||||
| * \ \ | merge 8.6 | dgp | 2017-12-06 | 1 | -0/+2 |
| |\ \ \ | |||||
| * | | | | Start RC branch for Tcl 8.6.8 | dgp | 2017-12-06 | 1 | -2/+2 |
| | | | | | |||||
* | | | | | Further fix for issue [ba1419303b4c]: Delete a namespace for an ensemble ↵ | pooryorick | 2017-12-20 | 1 | -1/+1 |
| | | | | | | | | | | | | | | | | | | | | | | | | | having a deletion trace deletes its namespace: segmentation fault. | ||||
* | | | | | Fix for issue [ba1419303b4c]: Delete a namespace for an ensemble having a ↵ | pooryorick | 2017-12-20 | 1 | -10/+7 |
| |_|_|/ |/| | | | | | | | | | | | | | | | deletion trace deletes its namespace: segmentation fault. | ||||
* | | | | [586e71dce4] Exeception handling at level #0 by EvalObjv | dgp | 2017-12-19 | 1 | -1/+1 |
| |_|/ |/| | | |||||
* | | | Plug memleak recently put into [package require]. | dgp | 2017-12-06 | 1 | -0/+4 |
|\ \ \ | |_|/ |/| | | |||||
| * | | Plug memleak recently put into [package require]. | dgp | 2017-12-06 | 1 | -0/+4 |
| | | | |||||
* | | | [0e4d88b650] Added enough refcounting to stop `make valgrind` complaints about | dgp | 2017-12-06 | 1 | -0/+2 |
| |/ |/| | | | | | "Invalid read". This is not a complete fix. Things are still broken. A working system of Namespace lifetime management looks like an 8.7 project. | ||||
* | | [ce3a211dcb] Failed file normalize when tail is empty string. | dgp | 2017-12-06 | 1 | -3/+2 |
|\ \ | |/ | |||||
| * | [ce3a211dcb] Failed file normalize when tail is empty string. | dgp | 2017-12-06 | 1 | -3/+2 |
| | | |||||
* | | [4f6a1ebd64] Stop crash when same value passed to the -map and -subcommands ↵ | dgp | 2017-12-05 | 1 | -154/+173 |
|\ \ | |/ | | | | | options of ensemble configuration. | ||||
| * | Factor clearing of ensemble subcommand table into utility routine. | dgp | 2017-12-05 | 1 | -29/+23 |
| | | |||||
| * | Another revised fix, much closer to sebres' patch now. | dgp | 2017-12-05 | 1 | -62/+90 |
| | | |||||
| * | [4f6a1ebd64] Different fix for the problem. Re-order the filling of the | dgp | 2017-12-04 | 1 | -59/+56 |
| | | | | | | | | | | | | subcommand table so there is no longer a conflict where multiple intreps of a single value are sought. If the mapDict and exportList are the same, then each key in the mapDict is known to be an element of the exportList without needing to check. | ||||
* | | [5808081213] Permit all bytearrays (including impure ones) to report length ↵ | dgp | 2017-11-30 | 1 | -4/+8 |
| | | | | | | | | without shimmering | ||||
* | | Don't provide the setFromAnyProc through the invalidRealType objType. This ↵ | jan.nijtmans | 2017-11-30 | 1 | -1/+1 |
| | | | | | | | | is a Tcl internal type, extensions shouldn't be able to convert their own Tcl_Obj to this. This shouldn't have been exposed to begin with. Tcl itself never calls it this way. | ||||
* | | Fix Tcl_UtfFindFirst()/Tcl_UtfFindLast(), which were broken by [83c0c569d6]. ↵ | jan.nijtmans | 2017-11-29 | 2 | -7/+61 |
| | | | | | | | | | | Not detected, because those functions aren't used anywhere in Tcl. So, added new test-cases, makeing sure this doesn't happen again. | ||||
* | | Update some functions in tclUtf.c to handle surrogate pairs when TCL_UTF_MAX ↵ | jan.nijtmans | 2017-11-29 | 1 | -16/+58 |
| | | | | | | | | | | == 4. Also update documentation to distinguish better between "Tcl_UniChar" and "Unicode character": Those are not necessary the same when TCL_UTF_MAX == 4. No change when TCL_UTF_MAX == 4 or TCL_UTF_MAX == 6. | ||||
* | | Change signature of (internal) TclScanElement() function. This saves memory ↵ | jan.nijtmans | 2017-11-15 | 5 | -32/+15 |
| | | | | | | | | allocation and the possibility for panic's in dict and list handling, requiring 1/4 of memory for internal allocation of temporary storage. No change to external API. | ||||
* | | [5d6de65036] [package require] with [package prefer stable] was not choosing ↵ | dgp | 2017-11-14 | 1 | -35/+58 |
|\ \ | |/ | | | | | available stable package. | ||||
| * | [5d6de65036] [package require] with [package prefer stable] was not choosing | dgp | 2017-11-14 | 1 | -33/+58 |
| |\ | | | | | | | available stable package. | ||||
| | * | Ticket [5d65e65036]. My fix. Do not skip the second check for stable ↵ | aku | 2017-11-13 | 1 | -33/+58 |
| | | | | | | | | | | | | versions even when the main check fails. Avoided radical changes to the structure (kept single search loop, with selection after). | ||||
* | | | No longer mark Tcl_EvalFile() as obsolete, since it will continue to be ↵ | jan.nijtmans | 2017-11-13 | 2 | -2/+0 |
|\ \ \ | |/ / | | | | | | | supported in Tcl 9.0 | ||||
| * | | No longer mark Tcl_EvalFile() as obsolete, since it will continue to be ↵ | jan.nijtmans | 2017-11-13 | 2 | -2/+0 |
| | | | | | | | | | | | | supported in Tcl 9.0 | ||||
* | | | Make "string split" and "string is (alpha|graph|...)" work as expected with ↵ | jan.nijtmans | 2017-11-10 | 1 | -3/+20 |
| | | | | | | | | | | | | | | | Unicode chars > U+ffff, when Tcl is compiled with TCL_UTF_MAX == 4. No effect when TCL_UTF_MAX == 3 or TCL_UTF_MAX == 6. Test-case added for "string split". | ||||
* | | | Make "scan %c" and the internal function ExtendUnicodeRepWithString() work ↵ | jan.nijtmans | 2017-11-09 | 2 | -5/+13 |
| | | | | | | | | | | | | as expected for TCL_UTF_MAX == 4 when handling characters > U+ffff. No effect when TCL_UTF_MAX == 3 or TCL_UTF_MAX == 6 | ||||
* | | | compiler warning | dgp | 2017-11-08 | 1 | -1/+1 |
|\ \ \ | |/ / | |||||
| * | | compiler warning | dgp | 2017-11-08 | 1 | -1/+1 |
| | | | |||||
* | | | [3298012] Stop RebuildTable asking ckalloc for more than it can give. | dgp | 2017-11-08 | 1 | -5/+8 |
|\ \ \ | |/ / | |||||
| * | | Attempted bug fix. | dgp | 2017-11-07 | 1 | -5/+8 |
| |/ | |||||
* | | TclOO object allocation: Set classPtr to NULL if it wasn't otherwise set. | dgp | 2017-11-08 | 1 | -1/+4 |
| | | |||||
* | | Fix [6f2f83cc149e9918884faffefebc8dfa695f4ea0|6f2f83cc14]: tclWinload.c ↵ | jan.nijtmans | 2017-11-03 | 1 | -0/+1 |
| | | | | | | | | robustness. And fix a minor possible memory leak in TclSetupEnv() as well. Thanks to Christian Werner for both suggestions, backported from Androwish. | ||||
* | | Better versions of TclGetIntFromObj and TclGetIntForIndexM macro's, which ↵ | jan.nijtmans | 2017-11-03 | 1 | -5/+13 |
| | | | | | | | | give advantage for platforms where longs are not ints (e.g. linux64) | ||||
* | | Cherrypick: [fc1409fc91] Method cloning needs to be careful with body ↵ | dkf | 2017-10-24 | 1 | -0/+1 |
| | | | | | | | | representations. | ||||
* | | Fix [b58e6897034fc5292c9d36ba8099d9a835c98172|b58e689703]: Return value of ↵ | jan.nijtmans | 2017-10-23 | 1 | -1/+1 |
| | | | | | | | | 'Tcl_Flush' | ||||
* | | Oops; put the code in the wrong place. Mixins have priority when deciding ↵ | dkf | 2017-10-19 | 1 | -12/+12 |
| | | | | | | | | method visibility. | ||||
* | | Ensure that method list introspection finds methods from mixins in all ↵ | dkf | 2017-10-19 | 1 | -19/+56 |
| | | | | | | | | cases. Use better approach for preventing excessive iteration. | ||||
* | | Typo's (Thanks to Gustaf Neumann), nothing functional. | jan.nijtmans | 2017-09-04 | 2 | -3/+3 |
| | | |||||
* | | [0e4d88b650] Allow command overwrite when deletion callback deletes namespace. | dgp | 2017-09-02 | 1 | -62/+104 |
|\ \ | |/ |