| 1 |
0.28 |
| 2 |
|
| 3 |
- Released at 2019-01-11T16:25:00+0900 |
| 4 |
- Now the distribution is made with mbtiny. |
| 5 |
- Various improvements of the test suite. |
| 6 |
- Rewrite the internally-used testing-purpose HTTP server to allow the control of 'Content-Length' request header. |
| 7 |
|
| 8 |
0.27: # 2016-10-28T12:59:00+0100 |
| 9 |
|
| 10 |
- Unbreak with Elasticeasrch 5.0. See https://rt.cpan.org/Public/Bug/Display.html?id=118425 |
| 11 |
|
| 12 |
0.26: # 2015-11-25T12:30:00+0100 |
| 13 |
|
| 14 |
- No functional changes since 0.25, but we had some Travis-specific |
| 15 |
changes in the repo, releasing just so we have the latest code there |
| 16 |
on the CPAN. |
| 17 |
|
| 18 |
0.25: # 2015-11-25T12:20:00+0100 |
| 19 |
|
| 20 |
- Make the t/select-timeout.t test which fails on various odd |
| 21 |
CPANtesters platforms a TODO. Maybe some OS-specific issue, maybe an |
| 22 |
issue with kill() in the CPANtesters sandboxes not behaving as we |
| 23 |
expect. |
| 24 |
|
| 25 |
0.24: # 2015-07-05T13:40:00+0200 |
| 26 |
|
| 27 |
- Minor copyediting and formatting changes to the documentation. No |
| 28 |
code changes at all. |
| 29 |
|
| 30 |
0.23: # 2015-07-03T17:00:00+0200 |
| 31 |
|
| 32 |
- The "Host" header can now be overriden by supplying a new |
| 33 |
`no_default_host_header` option along with a `Host` header in `head |
| 34 |
=> []` to request(). |
| 35 |
|
| 36 |
Before this we'd always send "Host: $host" over, where $host was the |
| 37 |
host we were connecting to, now you can customize this. |
| 38 |
|
| 39 |
- Fixed a bug where if passed passed `head => []` to request() we'd |
| 40 |
emit a ":" header, i.e. just an empty header name with an empty |
| 41 |
value. |
| 42 |
|
| 43 |
You could have just not passed the `head => ` value if the array was |
| 44 |
empty, but no we won't screw up and emit a single line consisting of |
| 45 |
":" if given an empty array. |
| 46 |
|
| 47 |
0.22: # 2015-05-27T07:54:17+0200 |
| 48 |
|
| 49 |
- No feature change. Re-package due to a missing file in the tarball: |
| 50 |
https://rt.cpan.org/Ticket/Display.html?id=104624 |
| 51 |
|
| 52 |
0.21: # 2015-05-22T15:26:23+0200 |
| 53 |
|
| 54 |
- Fix "Too many CRLF" issue. Hijk has been always generating HTTP |
| 55 |
request with an extra CRLF at the end. While many HTTP servers are |
| 56 |
ignoring those, some treat it as errors. We now eliminate the extra |
| 57 |
CRLF at the end of every request. |
| 58 |
See also http://www.w3.org/Protocols/rfc2616/rfc2616-sec4.html |
| 59 |
|
| 60 |
- Handle better when select() is interrupted by signals. |
| 61 |
|
| 62 |
0.20: # 2015-03-20-T15:10:00+0000 |
| 63 |
- Fix a regression in 0.14. When the chunked encoding support was |
| 64 |
introduced we accidentally stopped supporting "Content-Length: 0", |
| 65 |
treat those responses as the zero-body again. This broke e.g. PUT |
| 66 |
requests that would return no payload. |
| 67 |
|
| 68 |
- Add support for 204 No Content responses. According to the HTTP |
| 69 |
standard we must try to consume the body if there's no |
| 70 |
Content-Length, but not if the server returns a 204 response, then |
| 71 |
it MUST NOT include a body (see |
| 72 |
http://tools.ietf.org/html/rfc2616#page-60). |
| 73 |
|
| 74 |
This re-adds support for e.g. 204 No Content response, in practice |
| 75 |
this "worked" before 0.14, but only accidentally and other types of |
| 76 |
responses wouldn't work. |
| 77 |
|
| 78 |
- We now handle our system calls returning EINTR. |
| 79 |
|
| 80 |
0.19: # 2015-01-10-T18:30:00+0000 |
| 81 |
- Fix a major regression in 0.16. The introduction of "head_as_array" |
| 82 |
completely broke the disconnection logic when the socket_cache was |
| 83 |
enabled (which is the default). When talking to a webserver that |
| 84 |
would disconnect us after N requests request N+1 would always fail |
| 85 |
with a 0 byte response error. |
| 86 |
|
| 87 |
This issue was reported as RT #101424 |
| 88 |
(https://rt.cpan.org/Public/Bug/Display.html?id=101424) |
| 89 |
|
| 90 |
- Fix a minor regression in 0.16: The introduction of "head_as_array" |
| 91 |
broke the "proto" part of the return value in a relatively obscure |
| 92 |
edge case where we'd read the header & had no Content-Length and |
| 93 |
couldn't read() anything. |
| 94 |
|
| 95 |
- Fix an edge case in the Trailer support. It would only kick in if we |
| 96 |
got the Transfer-Encoding header before the "Trailer" header, not |
| 97 |
the other way around. |
| 98 |
|
| 99 |
0.18: # 2014-12-10T14:00:00+000 |
| 100 |
- We now do the right thing on "method => 'HEAD'". I.e. ignore the |
| 101 |
Content-Length parameter, previously we'd just hang trying to slurp |
| 102 |
up the body. |
| 103 |
- Fix an edge case with some of the live tests leaving around a HTTP |
| 104 |
server if they died, these don't run by default. |
| 105 |
|
| 106 |
0.17: # 2014-08-31T18:30:00+000 |
| 107 |
- Minor documentation changes, no functional changes. |
| 108 |
- The version number for the last release was incorrect in this |
| 109 |
changelog, fixed in this release. |
| 110 |
|
| 111 |
0.16: # 2014-08-31T00:10:00+000 |
| 112 |
- Major Change: There are several new Hijk::Error::* constants for |
| 113 |
common issues that happened during normal requests in the face of |
| 114 |
regular exceptions out of the control of this library, such as |
| 115 |
network blips. |
| 116 |
|
| 117 |
Existing code that checks $res->{error} should be |
| 118 |
forwards-compatible with this change, but anything that was doing |
| 119 |
e.g. regex checks against regular errors thrown by this library |
| 120 |
should be updated to check the new Hijk::Error::* constants instead. |
| 121 |
|
| 122 |
- It's now possible to specify "head_as_array" to get the returned |
| 123 |
headers as an array (with potential duplicated headers), rather than |
| 124 |
the default behavior of lossily returning them as a hash. |
| 125 |
|
| 126 |
- There's now a "read_length" option to control how much we |
| 127 |
POSIX::read($fd, $buf, $read_length) at a time. We don't expect this |
| 128 |
to be useful, it's mainly configurable on general principle so we |
| 129 |
don't have arbitrary unconfigurable hardcoded constants in the |
| 130 |
source. |
| 131 |
|
| 132 |
0.15: # 2014-08-30T10:00:00+000 |
| 133 |
- The new code to support chunked transfer encoding would return a |
| 134 |
nonexisting Hijk::Error::* value of "0" when it encountered a read |
| 135 |
timeout. This meant that not only was the error reporting broken, |
| 136 |
but anything checking if there were errors via the simple idiom of |
| 137 |
"if ($res->{error}) {...}" wouldn't properly report errors. |
| 138 |
|
| 139 |
We'll now correctly report these errors as |
| 140 |
Hijk::Error::READ_TIMEOUT. |
| 141 |
|
| 142 |
- Since there may still be other bugs like that in this new parsing |
| 143 |
mode it's disabled by default, if you know you want to parse chunked |
| 144 |
responses you have to pass parse_chunked => 1 for now. Usually you |
| 145 |
probably just want to disable chunked encoding on the other end, see |
| 146 |
the note about how to do that with nginx in the docs. |
| 147 |
|
| 148 |
0.14: # 2014-08-29T15:40:36+0900 |
| 149 |
- Start support chunked transfer encoding. |
| 150 |
|
| 151 |
0.13: # 2014-04-27T20:00:43+0200 |
| 152 |
- Switch to use non-blocknig fd to avoid a rare deadlock situation |
| 153 |
when select() is successful and the following read() blocks forever |
| 154 |
because there are really nothing to read. |
| 155 |
|
| 156 |
0.12: # 2014-01-31T18:20:00+0100 |
| 157 |
- Instead of dying on e.g. "Bad arg length for |
| 158 |
Socket::pack_sockaddr_in, length is 0, should be 4" when given a |
| 159 |
host we can't resolve we'll now return a $res with the error value |
| 160 |
set to Hijk::Error::CANNOT_RESOLVEif we can't gethostbyname() the |
| 161 |
provided hostname. Makes it easier to handle DNS resolution |
| 162 |
failures. |
| 163 |
|
| 164 |
0.11: # 2014-01-06T13:20:00+0100 |
| 165 |
- Fixed broken HTTP header parsing for servers that didn't return the |
| 166 |
entire header all at the same time, but in chunks. |
| 167 |
- We now return "proto" as well as "status" etc. in the response, so |
| 168 |
you can see what the protocol the server was using to speak to |
| 169 |
us. Also we pro-actively connections to servers that claim they're |
| 170 |
speaking HTTP/1.0. |
| 171 |
- Document that what the socket_cache is keyed on, for anyone wanting |
| 172 |
to implement a tied hash or whatever. |
| 173 |
- Fix a minor bug causing redundant work under "socket_cache => undef" |
| 174 |
|
| 175 |
0.10: # 2013-12-19T16:50:00+0100 |
| 176 |
- We can now talk HTTP/1.0 an addition to HTTP/1.1, have a way to |
| 177 |
disable the socket cache, and can specify connect and read timeouts |
| 178 |
independently. |
| 179 |
- Fix a really nasty bug with mixings up requests after encountering a |
| 180 |
timeout. See |
| 181 |
http://lists.unbit.it/pipermail/uwsgi/2013-December/006802.html for |
| 182 |
details. |
| 183 |
- Remove spurious requirenment on perl v5.14.2 |
| 184 |
- First stab at https://github.com/gugod/Hijk/issues/3 we'll now |
| 185 |
return an error key in the response with |
| 186 |
Hijk::Error::{CONNECT_TIMEOUT,READ_TIMEOUT} instead of dying. |
| 187 |
- Nuked the Hijk::HTTP::XS support from the repo, we've decided it was |
| 188 |
too complex for its own good. |
| 189 |
- Add support for an on_connect callback for seeing how long the |
| 190 |
connect/reads take. |
| 191 |
|
| 192 |
0.09: # 2013-12-13T07:38:25+0100 |
| 193 |
- KEEP CALM AND REMOVE FETCH OPTION |
| 194 |
- Hijk::request will use XS parser only if Hijk::HTTP::XS is loaded |
| 195 |
|
| 196 |
0.08: # 2013-12-12T20:10:00+0100 |
| 197 |
- We only checked for undefined return codes from POSIX::read(), not |
| 198 |
0, resulting in an infinite select/read loop when a server with |
| 199 |
keep-alive enabled cut off our connection. |
| 200 |
|
| 201 |
0.07: # 2013-12-09T12:50:00+0100 |
| 202 |
- Skip the live connect timeout test by default, it will fail making |
| 203 |
live connections on various firewalled/locked down hosts. |
| 204 |
|
| 205 |
0.06: # 2013-12-09T12:20:00+0100 |
| 206 |
- Declare missing test dependency on Test::Exception |
| 207 |
- Declare test dependency on Net::Ping 2.41 |
| 208 |
- Various POD improvements describing more limitations in the API and |
| 209 |
providing examples. |
| 210 |
- Don't unconditionally load the yet-to-be-released Hijk::HTTP::XS |
| 211 |
module, instead provide a "fetch" option. |
| 212 |
- Shutdown and delete the cached connection in case of read error. |
| 213 |
- Handle syswrite() returning undef without spewing an uninitialized |
| 214 |
comparison error |
| 215 |
- Various work on the test suite. |
| 216 |
|
| 217 |
0.05: # 2013-12-04T22:33:31+0100 |
| 218 |
- Properly invalidate connection cache when seeing 'Connection: close' in the response. |
| 219 |
|
| 220 |
0.04: # 2013-12-04T00:06:16+0100 |
| 221 |
- Implement 'connect timeout' and 'read timeout' |
| 222 |
|
| 223 |
0.02: # 2013-11-24T16:14:20+0100 |
| 224 |
- Passthrug extra HTTP header with the 'head' request arg. |
| 225 |
|
| 226 |
0.01: # 2013-11-24T01:49:08+0100 |
| 227 |
- Initial Release, with all wanted features are implemented. |
| 228 |
|