summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* fixes typo ("net" instead of outdated term "nett")sebres2019-05-244-10/+10
|
* timerate: be more precise using unsigned wide (for total execution time), ↵sebres2019-05-211-18/+21
| | | | avoid signed vs unsigned comparison warning (if compiled with VS).
* small amend (no cast needed)sebres2019-05-211-1/+1
|
* cherry-picing of [3f693cdfe9c875c4] to be more compatible across version, ↵sebres2019-05-212-5/+7
| | | | resolve mistake with "key objects/pointers comparison by the hash entry" firstly introduced there.
* comment typodgp2019-05-171-1/+1
|
* small amend to [ac566e9df84daeab] with rollback of done/continue optimizationsebres2019-05-162-0/+13
|
* timerate: allow continue from measurement cycle (used for conditional flow ↵sebres2019-05-162-15/+33
| | | | | | control of iterations); more gentle evaluation behaviour, similar to a cycle now (also avoids extra overhead to set result to interp, etc.); todo: rewrite optimization of INST_DONE using last-instruction pointer or optimization flags for compile of iteration (if sebres's perf-branch gets merged).
* DIGIT_BIT -> MP_DIGIT_BIT, since DIGIT_BIT is deprecated in future ↵jan.nijtmans2019-05-147-66/+108
| | | | | libtommath versions. So, better stop using it at all. Also add some other defines/typdefs for future libtommath compatibility.
* Revised documented advice that made obsolete reference to interp->result.dgp2019-05-111-4/+4
|
* (cherry-pick from8.6): Japanese Reiwa (令和) erajan.nijtmans2019-05-072-11/+13
|
* Plug memleak when deleting a namespace destroys a linked Tcl var.dgp2019-05-043-1/+53
|
* memleak demo testdgp2019-05-031-0/+11
|
* timerate: code style, doc styledkf2019-04-232-198/+356
|
* extend performance test-suite, allow several (repeatable) execution of ↵sebres2019-04-171-16/+62
| | | | | _test_run (if encosed in _test_start/_test_out_total) to produce same summary; provide possibility for measure of single iterators, etc. small code review
* Isolate tests of [info frame] results from testing environment.dgp2019-04-171-10/+10
|
* closes [940ce8f958] - tests/cmdMZ.test: avoid import timerate to global NS ↵sebres2019-04-091-0/+4
| | | | in tests (e. g. using tcltest -singleproc 1 -file 'cmdMZ* namespace*')
* closes [1e5e25cf2b] - tests/cmdMZ.test: fixed NRT-related sleeps (and ↵sebres2019-04-091-11/+19
| | | | | time-related corner cases and test expectations); todo: rewrite several tests if monotonic clock is provided resp. command "after" gets microsecond accuracy (RFE [fdfbd5e10] gets merged)
* Add test-cases for win32/win64 --disable-shared, and put standard ↵jan.nijtmans2019-04-081-2/+33
| | | | --enable-threads
* closes [1a3fa1232e306a44], test case cmdMZ-6.5 fixed to cover float value by ↵sebres2019-04-011-2/+5
| | | | iteration per second
* timerate: avoid divide by zero by no iterations in measurement cycle (e. g. ↵sebres2019-04-011-1/+11
| | | | count is 0)
* Update TZ info to tzdata2019ajan.nijtmans2019-03-276-55/+68
|
* Since only bytes 0xF0 - 0xF4 can be the first byte of a valid 4-byte UTF-8 ↵jan.nijtmans2019-03-241-3/+3
| | | | byte sequence, account for that in Tcl_UtfCharComplete(). Only effective when TCL_UTF_MAX>3
* Add entry for 元号 (or NewEra placeholder) to Unicode tables. Since Tcl ↵jan.nijtmans2019-03-212-16/+16
| | | | doesn't do rendering, this prepares Tcl for the expected may 1 event. See: [http://blog.unicode.org/2018/09/new-japanese-era.html]
* For Tcl >= 8.7, always compile-in the extended Unicode tables, no matter the ↵jan.nijtmans2019-03-172-8/+8
| | | | value of TCL_UTF_MAX. Do this in all Tcl versions, in order to prevent merge conflicts in future Unicode table updates.
* fixed mistake ($howmuch is substituted in tests and can be larger as last ↵sebres2019-03-081-1/+1
| | | | event index created by too small measurement time).
* merge sebres-8-5-timerate (TIP#527 - New measurement facilities in TCL: New ↵sebres2019-03-0516-194/+1931
|\ | | | | | | command timerate, performance test suite)
| * extended performance test-suite, since max-count is implemented in timerate, ↵sebres2019-03-052-95/+81
| | | | | | | | | | | | usage `::tclTestPerf::_test_run ?-no-result? reptime lst ?outcmd?`; update timer-event.perf.tcl for better readability (covering execution in multiple iterations now regarding max-count, so provides more precise result now); removed unused test-cases here (new cases of event-perf-branch only).
| * back-porting other performance test (timer-event.perf.tcl) from ↵sebres2019-03-051-0/+219
| | | | | | | | event-perf-branch
| * back-porting test-performance suite and clock.perf.tcl from clock-speedup branchsebres2019-03-052-0/+532
| |
| * helper to import `timerate` on demand in unknown/autoload proceeding.sebres2019-03-051-0/+3
| |
| * regarding the TIP#527, `timerate` shall be placed into `::tcl::unsupported` ↵sebres2019-03-051-7/+20
| | | | | | | | in versions prior to 8.7
| * highlightingsebres2019-03-051-4/+4
| |
| * amend: html-code order changedsebres2019-03-041-1/+1
| |
| * tools/tcltk-man2html: html-code for micro (sec) charactersebres2019-03-041-0/+1
| |
| * re-integrates the changes from the TIP#527 description into the manpagesebres2019-03-041-25/+37
| |
| * timerate documentation extendedsebres2019-02-131-4/+7
| |
| * timerate: extended with ?max-count? optional parameter, code review and more ↵sebres2019-02-133-16/+53
| | | | | | | | tests
| * small amend (correct wrong utf-8 prevention for micro sign in RE of check ↵sergey.brester2019-02-121-1/+1
| | | | | | | | test-case)
| * few test cases for timerate commandsergey.brester2019-02-121-0/+45
| |
| * fixes estimated time of too short execution considering calibrated overhead ↵sergey.brester2019-02-121-2/+2
| | | | | | | | (it is 0us and not 1us, example: `timerate {break}` with and without calibration)
| * timerate: allow break from measurement cycle (usable to provide conditional ↵sergey.brester2019-02-121-1/+8
| | | | | | | | stop possibility inside timerate)
| * cherrypick [8ad25ef9eb] from 8.6 - timerate: added dynamic factor by ↵sergey.brester2019-02-121-7/+24
| | | | | | | | threshold calculation (avoid growing of the execution time if iterations are not consistent, e. g. wax continuously on time)
| * merge 8.5 (note: to avoid too many conflicts, merged using `fossil merge ↵sergey.brester2019-02-12225-7970/+9599
| |\ | | | | | | | | | --baseline 0055a16a8b core-8-5-branch`)
| * | [win32] optimized calibration cycle (makes Tcl for windows "RTS" resp. ↵sebres2017-05-231-53/+122
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | NRT-capable): - the clock ticks never backwards (avoid it by negative drifts using comparison of times before and after calibration); - more precise, smooth/soft drifting (avoids too large drifts, already after 10 iterations the drift gets fewer as 0.1 microseconds); - because of more accurate drifting (aspire to the smallest difference), we can prolong calibration interval (up to 10 seconds by small tdiff-value); Closes ticket [b7b707a310ea42e9f1b29954ee8ca13ae91ccabe] "[win32] NRT-only - NativeGetTime backwards time-drifts bug"
| * | back-ported branch sebres-8-6-timerate (new command "timerate" for 8.5)sebres2017-05-1610-141/+944
| |\ \
* | | | Update internal tables to Unicode 12.0jan.nijtmans2019-03-012-974/+1011
| | | |
* | | | Move from trusty -> xenal for travis builds. Also add build on Mojave (xcode ↵jan.nijtmans2019-02-241-11/+11
| | | | | | | | | | | | | | | | 10.2)
* | | | No longer run socket test-cases on travis with OSX: There are too many ↵jan.nijtmans2019-02-241-3/+7
| |_|/ |/| | | | | | | | random failures/hangups.
* | | fixes [4ee9b4f3e965a7da5133]: result of command `time` overflows by single ↵sebres2019-01-281-1/+1
| | | | | | | | | | | | iteration longer as 35 minutes (uses wide-int instead of int as result now).
* | | Do not assume that literals have a non-NULL bytes field; user code could ↵dkf2019-01-251-15/+26
| | | | | | | | | | | | purge it.
generated by cgit v0.12 at 2024-11-17 09:21:36 (GMT)