| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
| |
* generic/tclEncoding.c (UtfToUnicodeProc, UnicodeToUtfProc):
* tests/encoding.test (encoding-16.1): fix alignment issues in
unicode <> utf conversion procs. [Bug 1122671]
|
|
|
|
|
| |
* tests/append.test(4.21-22): fix for longstanding [Bug 1570718],
lappending nothing to non-list. Reported by lvirden
|
|
|
|
| |
by dgp.
|
|
|
|
|
|
|
|
|
| |
* generic/tclInt.h:
* generic/tclIntDecls.h:
* generic/tclObj.c:
* generic/tclStubInit.c: added an internal function
TclObjBeingDeleted to provide info as to the reason for the loss
of an internal rep. [FR 1512138]
|
|
|
|
|
|
|
|
|
| |
* generic/tclHistory.c:
* generic/tclInt.h:
* generic/tclProc.c: made Tcl_RecordAndEvalObj not call "history"
if it has been redefined to an empty proc, in order to reduce the
noise when debugging [FR 1190441]. Moved TclCompileNoOp from
tclProc.c to tclCompile.c
|
|
|
|
|
|
|
| |
* tests/pkg.test: before comparison. The shorter string is the
smaller number. Added testcases as well. Interestingly all
existing test cases for vcompare compared numbers of the same
length with each other. See [SF Tcl Bug 1563836].
|
|
|
|
|
|
|
|
| |
* tests/expr.test: calculation done to determine whether a
shift in the (long int) type is possible. The calculation had
literal value "1" where it needed a value "1L" to compute the
correct result. Error detected via testing with the
math::bigfloat package [Bug 1567222]
|
|
|
|
| |
possible NULL derefs, [Bug 1566382] and coverity #33.
|
|
|
|
| |
{-1, 0, 1} to match expectations of CompareVersion() callers.
|
|
|
|
|
|
| |
* generic/regc_nfa.c (freearc): changed tests and asserts to
equivalent formulation, designed to avoid an explicit comparison
to NULL and satisfy coverity that 6 and 9 are not bugs.
|
|
|
|
| |
elements) code.
|
|
|
|
|
|
|
|
| |
* generic/regcomp.c (compile): the static function addmcce does
nothing when called with two NULL pointers; the only call is by
compile with two NULL pointers (regcomp.c #includes
regc_cvec.c). The whole thing is now ifdef'ed out with the macro
REGEXP_ADDMCCE_UNUSED. This also silences coverity's #7.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
boundary. [SF Tcl Bug 1563836].
* generic/tclPkg.c: [SF Tcl Bug 1563836]. Rewrote CompareVersion
to perform string comparison instead of numeric. This breaks
through the 32bit limit on version numbers. See code for details
(handling of leading zeros, signs, etc.). un-CONSTed some
arguments of CompareVersions, RequirementSatisfied, and
AllRequirementsSatisfied. The new compare modifies the string
(temporary string terminators). All callers use heap-allocated
ver-intreps, so we are good with that.
|
|
|
|
|
| |
TCL_GLOBMODE_TAILS and pathPrefix==NULL. This would cause a
segfault, as found by coverity #26.
|
| |
|
|
|
|
|
|
| |
demonstrates, "#define NULL 0" is just wrong, and as a quotable chat
figure observed, "If NULL isn't defined, we're not using a C compiler."
Improper fallback definition of NULL removed.
|
|
|
|
|
|
|
| |
required.
Tcl_Time structure members are longs. Support for varying compiler options
and build to platform-specific subdirs.
|
|
|
|
|
| |
aka coverity #51. Extended loop condition, added checking for
NULL to prevent seg.fault.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
0-length strings. This is coverity's bugs #54-5
|
|
|
|
| |
start==NULL and numBytes<0. This is coverity's bug #20
|
|
|
|
| |
end of the structure, for better backward compatibility.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* generic/tclDecls.h: Regenerated from tcl.decls.
* generic/tclStubInit.c:
* doc/PkgRequire.3: Documentation of extended API,
* doc/package.n: extended testsuite.
* tests/pkg.test:
* generic/tcl.decls: Implementation.
* generic/tclBasic.c:
* generic/tclConfig.c:
* generic/tclInt.h:
* generic/tclPkg.c:
* generic/tclTest.c:
* generic/tclTomMathInterface.c:
* library/init.tcl:
* library/package.tcl:
* library/tm.tcl:
|
| |
|
|
|
|
| |
TclParseNumber covers the task just fine.
|
| |
|
|
|
|
|
|
|
| |
* tests/expr.test: was completely broken. Falling back on use of
bignums for all non-trivial ** calculations until
native-type-constrained special cases can be done carefully and
correctly. [Bug 1561260].
|
|
|
|
| |
* tests/expr.test: $x**1 as $x**3. [Bug 1555371]
|
|
|
|
| |
no longer needed.
|
|
|
|
|
|
| |
disabling the INST_TRY_CVT_TO_NUMERIC instruction at the end
of an expression when function arguments contain operators.
[Bug 1541274]
|
| |
|
| |
|
|
|
|
|
| |
* generic/tclPkg.c (Tcl_PkgRequireEx): replace Tcl_GlobalEval(Obj)
with more efficient Tcl_Eval(Obj)Ex
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
orphan ExprNodes to store the closure of left pointers. This
lets us avoid repeated re-scanning leftward for the left
boundary of subexpressions, which in worst case led to near
O(N^2) runtime.
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
* generic/tclParseExpr.c: to copy only whole characters when
doing %s formatting. This relieves callers of TclObjPrintf() and
TclFormatToErrorInfo() from needing to fix arguments to character
boundaries. Tcl_ParseExpr() simplified by taking advantage.
[Bug 1547786]
|
|
|
|
|
| |
to count up the number of arguments required by examining the
format string. [Bug 1547681]
|
| |
|
| |
|
| |
|