Debian Codegraph @

ChangeLog

raw
1 2012-05-19: version 1.30
2
3 * Fix buck2sock corruption that causes errors like
4 "No map found matching for GLOB" or "No sock found for" ([email protected])
5 https://rt.cpan.org/Ticket/Display.html?id=62872
6
7 * Fix t/05_reconnect_timeout.t to not fail on networks that fast-reject
8 TEST-NET-1 https://rt.cpan.org/Ticket/Display.html?id=74500
9
10 * Fix warning about undefined value in GetParser.pm ([email protected])
11 https://rt.cpan.org/Ticket/Display.html?id=31067
12
13 * Add documentation of connect_timeout and select_timeout
14 https://rt.cpan.org/Ticket/Display.html?id=30904
15
16 2010-06-17: version 1.29
17
18 * fix warnings spew if calling stats on a down server
19
20 * fix buck2sock confusion (Eddie Canales)
21
22 * quell ipv6-related warning
23 https://rt.cpan.org/Ticket/Display.html?id=51761
24
25 2009-10-21: version 1.28
26
27 * IPv6 support (https://rt.cpan.org/Ticket/Display.html?id=50577)
28
29 * Add Encode.pm requirement to Makefile.PL for perl 5.6.x (Ask).
30
31 2009-09-22: version 1.27
32
33 * Fix get() with utf-8 keys (athomason)
34
35 * "stats malloc" command is gone in 1.4; remove those tests (athomason)
36
37 * Add append/prepend support (dormando)
38
39 * Fix occasional failure in the 100_flush_bug.t test (Ask Bjørn Hansen)
40
41 2009-05-04: version 1.26
42
43 * don't include "stats sizes" by default in the stats method,
44 as that can hang big servers for a few seconds (Brad Fitzpatrick)
45
46 2009-05-02: version 1.25
47
48 * Clear @buck2sock when calling disconnect_all. (Dennis Stosberg,
49 [rt.cpan.org #45560]
50
51 * Reconnects to a dead connection shouldn't happen every time when the
52 connection has never succeded. Apply the dead timeout to sockets that
53 never even came up. Add a test.
54
55 * Warn when trying to put undef values into memcache.
56 (Henry Lyne <[email protected]>)
57
58 * flush_all now only returns success if there is a proper reply from all
59 servers - Yann Kerherve <[email protected]>
60
61 * 'noreply' support from Tomash Brechko <[email protected]>
62
63 * various test updates from Ronald J Kimball <[email protected]>
64
65 2007-07-17: version 1.24
66
67 * update the stats method, including tests for it
68 (Ronald J Kimball <[email protected]>)
69
70 * arguments to 'new' constructor can be %hash or $hashref now
71 (previously was only $hashref)
72
73 * work around a Perl segfault (Matthieu PATOU <[email protected]>)
74 see http://lists.danga.com/pipermail/memcached/2007-June/004511.html
75
76 2007-06-19: version 1.23
77
78 * add 'remove' as an alias for 'delete' (Dave Cardwell <[email protected]>)
79
80 2007-06-18: version 1.22
81
82 * lost connection handling broken due to wrong %sock_map indexing
83 http://rt.cpan.org/Public/Bug/Display.html?id=27181
84 fix from RHESA
85
86 * let parser_class be configured as a constructor option,
87 defaulting to XS if available, else regular. (unless
88 $ENV{NO_XS} is set, in which case the default is regular)
89
90 2007-05-02: version 1.21
91
92 * new faster optional interface for GetParser subclasses. doing
93 this release so upcoming Cache::Memcached::GetParserXS can
94 depend on this. otherwise this release isn't interesting.
95
96 2007-04-16: version 1.20
97
98 * fix "Warning produced when flush_all called" from CDENT
99 http://rt.cpan.org/Public/Bug/Display.html?id=22181
100
101 * support access via unix domain sockets. (Nathan Neulinger <[email protected]>)
102
103 * abstract out response parsing into own class, and add XS-module
104 detection, so if you have the XS (C) version, things'll be faster.
105 that part's not done yet.
106
107 2006-07-03
108 * don't use dual scalar/glob sockets. makes it all profilable
109 again under SmallProf, DProf, and Devel::Profiler, all three
110 of which used to barf on those weird sockets previously
111 * only init_buckets once, when servers are changed
112 * don't call sock_to_host and get_sock as much: cache closer
113 in get_multi
114 * more internal caching (buck2sock, etc)
115 * fast paths for namespaces/single sock/etc in a few more places
116 * general micro-speedups all over
117
118 2006-06-27
119 * patch from Maxim Dounin <[email protected]> to fix a typo
120 which caused no_rehash flag to not work.
121 * release 1.18
122
123 2006-04-29
124 * flush_all command from Patrick Michael Kane <[email protected]>
125 * document namespaces
126 * release 1.17
127
128 2006-04-29
129 * fix stats method (people have only been asking since 2004 :-/)
130 * add tests
131 * move Memcached to lib/Cache directory to be more CPAN-friendly
132 * release 1.16
133
134 2005-09-20
135 * configurable connect delays and callback on connect failure (brad)
136 * release 1.15
137
138 2005-08-09
139 * _connect_sock never works in blocking mode because of a bug in setting
140 the default timeout. (Gaal)
141
142 2004-07-27
143 * release 1.14
144
145 2004-07-27
146 * kill buggy, slow ord() _hashfunc, replace with crc32.
147 this adds String::CRC32 as a dependency. thanks to
148 everybody's feedback on the mailing list.
149
150 2004-07-19
151 * don't use pos() because it doesn't seem to work in
152 taint mode. use $+[0] instead. (Dave Evans <[email protected]>)
153
154 2004-06-19
155 * support multiple paths to memcache nodes (Brad)
156 see 'set_pref_ip'
157
158 2004-05-30
159 * release version 1.13
160
161 2004-05-26 (Whitaker <[email protected]>)
162 * quiet warning
163
164 2004-05-25 (Whitaker <[email protected]>)
165 * get_multi shouldn't modify caller's @_
166
167 2004-05-18 (Michael <[email protected]>)
168 * namespace support
169 * use fields
170
171 2004-05-16 (Alexei Kozlov <[email protected]>)
172 * remove warnings with vec
173
174 2004-04-09 (brad)
175 * in perl 5.6, trap errors dethawing 5.8 storable objects
176 and instead treat it like a cache miss
177
178 2004-04-01
179 * use $! and not %! for perl 5.6 compat (Dave Evans <[email protected]>)
180 * don't mark whole IP dead anymore when a node is down (Jason Titus <[email protected]>)
181 * start version numbering (Jamie McCarthy <[email protected]>)
182
183 2004-03-09 (Brad/Avva)
184 * _oneline can return more than one line (but always on a line break),
185 so caller must decide when it's really time to quit. had to modify
186 run_command to know that. (which is used by stats)
187
188 2004-03-05 (Dave Evans <[email protected]>)
189 * Here's a really trivial patch for the Perl binding,
190 Cache::Memcached. The bug is that the module assumes that the
191 currently select()ed filehandle is STDOUT, but this might not be
192 the case. So this patch ensures that the select()ed filehandle is
193 preserved, not forced to STDOUT.
194
195 2004-02-29 (Brad)
196 * add readonly option
197
198 2004-02-27 (Avva)
199 * Cleaner handling of the case when _oneline is called without a
200 line parameter (i.e. not to send anything, just read a line from
201 the socket). Make it depend on $line being defined only,
202 regardless of its content (thanks Brad!).
203
204 2004-02-25 (Avva)
205 * Asyncify all I/O, finally get rid of alarm() yuckiness, unify all
206 one-liner command/responses into a single internal API.
207
208 2004-02-17
209 * document in POD the delete method
210
211 2004-02-03
212 * fix bug with 2k read boundaries falling in the middle
213 of "VALUE ..." or "END" lines, thus halting future
214 parsing and responses. (eek!)
215 * version 1.0.12
216
217 2003-12-01
218 * merge stats/stats_reset patch from Jamie McCarthy
219 * trailing whitespace cleanup
220
221 2003-11-08
222 * work on Solaris/BSD where there's no MSG_NOSIGNAL.
223 the expense is extra syscalls to change the local
224 SIGPIPE handler all the time. in the future, it'd
225 be nice to have an option so Solaris/BSD callers
226 can say, "Hey, I've turned off SIGPIPE globally,
227 don't worry about it."
228
229 2003-10-26
230 * add a test file, so automated CPAN test hosts are happy
231 * check MSG_NOSIGNAL immediately on module load, not on use,
232 so Solaris dies early. (still on TODO to fix, but better
233 to fail loudly)
234 * version 1.0.11
235
236 2003-10-25
237 * version 1.0.10, rename to Cache::Memcached, upload to CPAN
238
239 2003-10-18
240 * implement read/write timeouts everywhere. Now the client shouldn't
241 hang if the server machine goes down unexpectedly. (avva)
242
243 2003-10-16
244 * use Storable::nfreeze instead of freeze, so hosts from different
245 architectures can all use the same data. (all must use Perl, though.
246 the different memcache APIs all store/pickle/serialize data differently)
247 Suggestion by Jason Titus <[email protected]>
248
249 2003-10-06
250 * fix _incrdecr to return complete number, not just first
251 digit (thanks to Ryan T. Dean)
252 * release version 1.0.9
253
254 2003-10-04
255 * document expiration times in POD (thanks to Tim Bunce
256 for noting the omission)
257 * release version 1.0.8
258
259 2003-10-03
260 * add connect timeout of 0.25s, for dead host detection.
261 We had 1 second a couple revs ago, but lost it when
262 ditching IO::Socket module. (avva)
263
264 2003-10-02
265 * fix _incrdecr with explicit-hashvalue keys (whitaker)
266
267 2003-10-01
268 * add run_command API call. TODO: document, and document
269 the $exptime on the setters
270
271 2003-09-30
272 * use send instead of print, so we can set MSG_NOSIGNAL
273 and not get SIGPIPES, which avoids 3 syscalls of localizing
274 $SIG{PIPE} and sends everything at once, instead of 4k
275 stdio chunks. in review: stdio buffered in, send unbuffered
276 out. TODO: setvbuf so reads are buffered at more than 4k.
277
278 2003-09-29
279 * yet faster parsing
280 * switch to stdio/perlio instead of raw io: more correct,
281 simpler parsing code.
282
283 2003-09-28
284 * prevent some warnings
285 * faster get() call that doesn't use get_multi()
286 * optimizations for single-server case
287 * use socket APIs directly, instead of uber-slow IO::* modules
288 * new faster _load_items parsing
289
290 2003-09-04
291 * emit debug when set/add/replace fails, in addition to succeed
292
293 Version 1.0.7
294 -- compression support (Brad Whitaker)
295
296 Version 1.0.6
297 -- incr/decr client support
298 -- make delete optionally take second argument (server now supports
299 a delay time on delete)
300 -- doc updates from Jamie McCarthy
301 -- better hashing after dead host detection: new requests go to different
302 remaining hosts, instead of all to the same one.
303
304 Version 1.0.2
305 -- initial release, about.