| 1 | Revision history for Perl extension Convert::UUlib. |
1 | Revision history for Perl extension Convert::UUlib. |
| 2 | |
2 | |
| 3 | _FP_gets still 70% of scan time. decode time: 24% _FP_gets, uulib_crc32 23%, UUValidData 11% |
3 | TODO: biggest timesinks: FP_fgets/getc_unlocked overall, UUInsertPartToList dominates large loads due to O(n**2) search |
| 4 | |
4 | |
|
|
5 | 1.8 Thu Dec 17 02:23:53 CET 2020 |
|
|
6 | - no bugfixes in this release due to lack of known bugs, but the major changes |
|
|
7 | in this release might have introduced new bugs, so watch out. |
|
|
8 | - update large decoder example to disable OPT_AUTOCHK and use Smerge -1. |
|
|
9 | - some micro-optimisations. |
|
|
10 | - avoid costly string comparisons by comparing hashes when isnerting items, |
|
|
11 | which speeds up insertion by a constant factor. |
|
|
12 | - improve uulist (and other) structure layout and size. |
|
|
13 | - reverse the order of file list items, which heuristically improves |
|
|
14 | match speed with large usenet file lists. |
|
|
15 | - use getc instead of fgetc, which makes no difference on gnu/linux. |
|
|
16 | - clean up _FP symbol names toi not start with an underscore. |
|
|
17 | - use feof_unlocked and ferror_unlocked. |
|
|
18 | - implement a faster ascii-only strnicmp. |
|
|
19 | - misc very minor code improvements. |
|
|
20 | - remove quite a bit of pre-posix/dos/etc. cruft. |
|
|
21 | - use flockfile, if available, for a potential but small |
|
|
22 | speed gain. |
|
|
23 | - speed up inner yenc decoder loop. |
|
|
24 | - kentnl said that this module should no longer claim to be a simple |
|
|
25 | interface to uulib, as the bunndled copy is now better maintained |
|
|
26 | than the original upstream library. |
|
|
27 | |
|
|
28 | 1.71 Tue Mar 17 00:54:06 CET 2020 |
|
|
29 | - backport to c89 (patch by Paul Howarth). |
|
|
30 | |
|
|
31 | 1.7 Sat Feb 29 22:07:54 CET 2020 |
| 5 | - new function: GetFileList. |
32 | - new function: GetFileList. |
| 6 | - experimental perlmulticore support (see manpage). |
33 | - experimental perlmulticore support (see manpage). |
|
|
34 | - Initialize is now a NOP and CleanUp automatically initializes again. |
| 7 | - updated example decoder a bit. |
35 | - updated example decoder and documentation a bit. |
| 8 | - include ecb.h to deal with compiler builtins and endianness. |
36 | - include ecb.h to deal with compiler builtins and endianness. |
| 9 | - some further µ-optimisations in hot code, especially for yEnc. |
37 | - some further µ-optimisations in hot code, especially for yEnc. |
| 10 | - replace crc32 function by slice-by-16 version by Stephan Brumme, |
38 | - replace crc32 function by slice-by-16 version by Stephan Brumme, |
| 11 | which should speed up yEnc en-/decoding. |
39 | which should speed up yEnc en-/decoding. |
| 12 | - yEnc: do not calculate two crcs per part, instead, combine |
40 | - yEnc: do not calculate two crcs per part, instead, combine |
| … | |
… | |
| 14 | - yEnc: allow pcrc= in addition to pcrc32= for yenc trailers, as |
42 | - yEnc: allow pcrc= in addition to pcrc32= for yenc trailers, as |
| 15 | some draft mentions both and it is actually in active use. |
43 | some draft mentions both and it is actually in active use. |
| 16 | - yEnc: ignore crc32= on multiparts, except on the last part, |
44 | - yEnc: ignore crc32= on multiparts, except on the last part, |
| 17 | which avoids spurious corruption warnings. |
45 | which avoids spurious corruption warnings. |
| 18 | - be more precise in documenting code licenses in COPYING. |
46 | - be more precise in documenting code licenses in COPYING. |
|
|
47 | - convert constant creation to the method I normally use. |
|
|
48 | - use common::sense. |
| 19 | |
49 | |
| 20 | 1.62 Mon Feb 17 23:19:42 CET 2020 |
50 | 1.62 Mon Feb 17 23:19:42 CET 2020 |
| 21 | - major performance improvement by simplifying code in _FP_gets |
51 | - major performance improvement by simplifying code in _FP_gets |
| 22 | to not use fscanf. This might slow things down on platforms |
52 | to not use fscanf. This might slow things down on platforms |
| 23 | with very slow fgetc. |
53 | with very slow fgetc. |
