Jim Tcl: Timeline

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

20 most recent timeline items

2013-03-30
01:33		[fde1b605ed] Leaf: Makefile.in: Depend on the headers Removes the need to do a "make clean" after changing jim.h or other headers. Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
2013-03-27
22:44		[e9e44f9914] Properly handle systems where (-ve % +ve) returns +ve Can't just use abs() This is the correct fix rather than c7e5c48 Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
21:19		[84bc933282] Update autosetup to v0.6.5 Includes a fix for -gstabs on newer macs Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
21:17		[8058474b6f] Need abs() since modulo of -ve is impl dependent Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
03:44		[60f3f4513d] More small improvements to JimWideToString Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
02:20		[9f23823583] Further improvements to JimWideToString Small time and space improvements Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
01:35		[e9aee28cda] Implement WideToString locally, printf is slow Even though it shouldn't need to, sprintf uses the libc-internal mutex, locale support, and other parts making it slow. [while] busy loop - 306 ms 311 ms - [for] busy loop - 180 ms 182 ms - mini loops - 146 ms 146 ms - fibonacci(25) - 191 ms 176 ms 92.1466 heapsort - 138 ms 131 ms 94.9275 sieve - 182 ms 143 ms 78.5714 sieve [dict] - 162 ms 138 ms 85.1852 ary - 188 ms 166 ms 88.2979 ary [dict] - 173 ms 156 ms 90.1734 repeat - 132 ms 130 ms - upvar - 142 ms 147 ms - nested loops - 173 ms 173 ms - rotate - 22 ms 22 ms - dynamic code - 108 ms 97 ms 89.8148 dynamic code (list) - 45 ms 46 ms - PI digits - 333 ms 319 ms 95.7958 expand - 46 ms 45 ms - wiki.tcl.tk/8566 - 251 ms 235 ms 93.6255 mandel - 212 ms 199 ms 93.8679 Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
01:04		[7a3c3748fb] Remove most quotingType allocations in JimMakeListStringRep Only two bits are needed, so also change to unsigned char from int. Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
2013-03-26
12:44		[40f36dbef9] Avoid list allocations under four pointers (16/32 bytes) It's more overhead than usable space to get space for two pointers. Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
2013-03-25
13:17		[224b7724c6] Put short-lived hash iterator allocations on the stack No reason to have these on the heap for such short whiles. They are small (32 bytes) and only used within the same function. v2: The external interface is preserved. Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
2013-03-24
22:35		[dff9f4bef9] Shuffle ScanFmtPartDescr to reduce size (40 -> 32) Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
2013-03-14
00:10		[2aff467da8] Trivial optimisation Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
2013-03-13
01:46		[28ecda72d4] Shuffle ExprByteCode to reduce size on 64-bit (24->16) Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
01:46		[3e089f33d1] Make expr array more compact Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
2013-03-11
23:00		[e2546815df] Shuffle Jim_Obj to shrink it on 64bit (72 -> 64), removing holes No perf difference measured using bench.tcl. Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
23:00		[93cd0d7b6e] Precompute operator length for JimParseExprOperator, optimize logic No regressions, no changes in bench.tcl. JimParseExprOperator was shockingly inefficient, to the point of taking 63% of cpu in my own app. It did a linear search over >60 structs, re-computing the string length each time, and doing a string comparison for each. This patch pre-computes the operator string length, speeding the function up by two. Adding the first-letter fast path and evaluating length before strncmp make for another 2x speedup, making the total speedup 4x. Signed-off-by: Lauri Kasanen <cand@gmx.com> (user: steveb@workware.net.au, tags: trunk)
2013-02-15
02:26		[cfca969b62] Fix [string replace] replacing single char Also added some additional tests Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)
2012-12-14
04:30		[b21543be23] Update linenoise.c to match recent github From git://github.com/msteveb/linenoise.git Allow tab-completion only at the end of line Replace magic number 9 with '\t' in linenoisePrompt() Fix several warnings from gcc. Fix first-chance exceptions in Windows - WriteConsoleOutputCharacter() didn't have its final parameter set Add MSVC support - Test for _WIN32 to check for building for Windows. (user: steveb@workware.net.au, tags: trunk)
2012-11-29
00:30		[5494d1bea4] Use sqlite3_prepare_v2() When I'm working with SQLite3 databases in Jim, I find that I occasionally get a 'schema has changed' error so I end up using catch to detect the error and try the query again. Here's a quick patch to change the SQLite3 extension to use the sqlite3_prepare_v2() API function instead of sqlite3_prepare() - trivial but the _v2() function has a number of benefits and everyone should have a new enough SQLite3 library to support it. Please can you consider it for inclusion? See - http://www.sqlite.org/c3ref/prepare.html The benefit I'm specifically interested in is: With the _v2() variant, "If the database schema changes, instead of returning SQLITE_SCHEMA as it always used to do, sqlite3_step() will automatically recompile the SQL statement and try to run it again." (user: steveb@workware.net.au, tags: trunk)
2012-10-31
10:09		[171c4f06ce] Fix linenoise serial console window size for vt102 This includes minicom, which doesn't support hpa (CHA), only cuf. Signed-off-by: Steve Bennett <steveb@workware.net.au> (user: steveb@workware.net.au, tags: trunk)