HoneyryderChuck
c7f177adbb
added regexp extensions where using match?
2021-08-10 11:25:36 +01:00
HoneyryderChuck
e19041aaad
no ned to call Kernel directly (at least, not anymore)
2021-08-10 11:10:45 +01:00
HoneyryderChuck
c4f5362156
making responses coercable to string
2021-08-10 11:10:45 +01:00
HoneyryderChuck
20db082db2
changing proc to lambda (just because of rbs)
2021-08-10 11:10:45 +01:00
HoneyryderChuck
802c47a037
multipart decoding: fail if no boundary found in header
2021-08-10 11:10:45 +01:00
HoneyryderChuck
e1ee8c69dc
proxy: fixing proxy resolve error filtering to also work with system resolver
2021-08-10 11:10:45 +01:00
HoneyryderChuck
6b61b8ccdb
fixing signatures
...
also adding some checks on code, in order for steep to stop complaining
about potential nil returns.
2021-08-10 10:28:58 +01:00
HoneyryderChuck
556c94a575
added multipart decooder, which overtakes on Response#form for multipart payloads
2021-08-09 15:54:25 +01:00
HoneyryderChuck
f2d3c1f09b
added Response#form (supports only x-www-urlencoded for now)
2021-08-09 15:54:25 +01:00
HoneyryderChuck
a85828d0d5
added Response#json
2021-08-09 15:54:24 +01:00
HoneyryderChuck
c9de63d4cd
bumped version to 0.16.1
2021-08-09 15:54:16 +01:00
HoneyryderChuck
582447f10e
bugfix: fixed native resolver timeout when dealing with absolute DNS
...
names
DNS answers are decoded using `resolv` internal structures. In this
particular case, we were using `Resolv::DNS::Name#to_s` to infer the
query dns name an answer refers to. This failed, because, as per
documentaation:
```ruby
p Resolv::DNS::Name.create("x.y.z.").to_s #=> "x.y.z"
p Resolv::DNS::Name.create("x.y.z").to_s #=> "x.y.z"
```
this caused issues when marking the DNS resolving as done. As a fix, I
applied the same logic from `Resolv::DNS::Name#inspect`. which correctly
prints absolute names.
Fixes #145
2021-08-09 13:36:49 +01:00
HoneyryderChuck
f9c17af6f3
save against file handle from system resolve being closed abruptly
2021-08-05 23:29:49 +01:00
HoneyryderChuck
48a2d82c34
http2 connection: removed interests calculation traversing requests
...
the traversal seems unnnecessary, given the subsequent cheaper checks. A
request with a stream will always be writable unless it's done, or is
expecting a 1xx response; however, there'll be drains in place, and
that's a cheaper check.
2021-08-05 22:52:58 +01:00
HoneyryderChuck
59755d7d01
connection: avoid calculating interests when there was nothing that could have changed the result
2021-08-05 22:51:16 +01:00
HoneyryderChuck
63c39a4032
fixed respond_to_missing? sig
2021-08-05 15:03:50 +01:00
HoneyryderChuck
65261217b1
optimization for calculation whether parser is empty
...
This avoids traversing all requests in the parser. this is possible for
HTTP/1.1 because, given that requests are sequentially concluded, one
can get away with checking only first and last request state.
2021-08-05 15:03:49 +01:00
HoneyryderChuck
3d549817cb
optimization on build_request to reduce Options object allocations on
...
multi-request calls
request options already get aggregated in #build_requests. this
optimization removes the Options object rebuild for each request,
thereby sharing the same across concurrent requests.
this is possible because, although #build_request is public, we document
it only with passing hash of options, never an Options object. This
would probably result in errors.
2021-08-05 15:03:49 +01:00
HoneyryderChuck
cc02679b80
Options#to_hash: avoid intermediate array, directly assign to final hash
2021-08-05 15:03:49 +01:00
HoneyryderChuck
74f355314d
removing reliance on regexp ops in http1 parser
2021-08-05 15:03:49 +01:00
HoneyryderChuck
29105854e3
do not pass around options after requests get initialized, instead rely on rquest options
2021-08-05 15:03:49 +01:00
HoneyryderChuck
b126938a65
skip options merge if the options to merge are already included
2021-08-04 16:30:20 +01:00
HoneyryderChuck
d23be82b7e
bumping versio to 0.16.0
2021-08-04 00:16:28 +01:00
HoneyryderChuck
953f2d0d13
fixed calculation of connection header when explicit connection header
...
is passed
also, fixed the test which was relying on the previous one-time-to_s
behaviour of responnse bodies.
2021-08-01 00:38:40 +01:00
HoneyryderChuck
ba742847f7
fixing grpc dependency on body being closed when buffered
2021-07-24 04:00:19 +01:00
HoneyryderChuck
a1be10cbc6
defining finalizer on response body, which keeps the reference to the buffer, which will get closed on GC; this will ensure that file descriptors don't leak
2021-07-24 04:00:19 +01:00
HoneyryderChuck
e43d5eddcd
making body comparison less wasteful, while keeping existing poinnters after call to to_s
2021-07-24 04:00:19 +01:00
HoneyryderChuck
0f1d8cb271
make Response#to_s non-destructive, and keep the body around
...
for most cases, this didn't make sense, i.e. response bodies were so
small, and the user is fine with keeping it all in mem, whereas
clearing the string is more confusing. And it breaks pattern
matching when matching body multiple times.
2021-07-24 04:00:19 +01:00
HoneyryderChuck
7aea7fe471
removing a few method_missing implementations
...
replacing them with SimpleDelegator where it made sense, and removing
them altogether whenever possible. ErrorResponse doesn't have a special
handling for Response methods anymore, but it's highly debatable if it
should.
2021-07-21 15:31:49 +01:00
HoneyryderChuck
15143f7322
went more meta with grpc and defining rpcs with def, optimal lookup up, removing method_missing logic
2021-07-21 15:31:18 +01:00
HoneyryderChuck
a9270a7616
do not require protobuf, as the user might use a different protobuf lib
2021-07-20 17:29:51 +01:00
HoneyryderChuck
e0a46dc7ec
making the grpc plugin build a rpc definitions from ma grpc generic service, thereby diminishing boilerplate
2021-07-20 17:29:03 +01:00
HoneyryderChuck
37f23ad8c3
cookie: force name to string; as for the value, only late-force it on writing the cookie
2021-07-16 09:52:21 +01:00
HoneyryderChuck
00a6e4672d
make sure that cookie jar on request is used, in case request has a distinct set of options
2021-07-16 09:52:21 +01:00
HoneyryderChuck
3ae18120d2
when users pass a Cookie header as option and plugin is enabled, the header will be parsed and managed by the jar
2021-07-16 09:52:21 +01:00
HoneyryderChuck
ac398ae0f0
bug (unexisting var)
2021-07-15 15:53:15 +01:00
HoneyryderChuck
174f5f9647
allow cookie jar to merge
2021-07-15 15:53:15 +01:00
HoneyryderChuck
1a7025941b
allow any option which can be merged, to be merged. This is a semi-breaking change, as some hash-like options will expanded instead of replaced, like the resolver options, but that was unspecified before, no guaarantees
2021-07-15 15:53:15 +01:00
HoneyryderChuck
2ea8b2ec2d
Merge branch 'issue-135' into 'master'
...
Options
Closes #135
See merge request honeyryderchuck/httpx!148
2021-07-14 17:31:11 +00:00
HoneyryderChuck
ac39eb8d36
bump version to 0.15.4
2021-07-14 09:52:39 +01:00
HoneyryderChuck
514e6cafb3
do not call .marshal directly
2021-07-13 21:12:53 +01:00
HoneyryderChuck
e9e3063dcc
fixing options class hierarchy, which caused truffleruby to fail
2021-07-13 19:56:59 +01:00
HoneyryderChuck
cab5a94d31
removing setters from Options, which weren't public API anyway
2021-07-13 19:56:59 +01:00
HoneyryderChuck
9d910df01b
fixing order of sub-plugins load in nnthe aws sigv4 plugin
2021-07-13 19:56:59 +01:00
HoneyryderChuck
5604501d01
using the new OptionsMethods way for defining internal plugin options
2021-07-13 19:56:59 +01:00
HoneyryderChuck
5c2ad6b588
allow definition of options via OptionsMethods module; using .extra_options for setting defaults
2021-07-13 19:56:59 +01:00
HoneyryderChuck
ec0f3cd163
defining options via normal method definition, def option_optname
2021-07-13 19:56:59 +01:00
HoneyryderChuck
340fb6c7c6
Merge branch 'ojab-master-patch-08963' into 'master'
...
Don't count `nil` as an retryable error class
See merge request honeyryderchuck/httpx!147
2021-07-11 15:18:53 +00:00
ojab
246094d8fc
Remove uneeded if defined?(TLSError) check
...
It's always defined
2021-07-11 15:05:21 +00:00
ojab
b5fe5bec11
Remove unused method definition, it's defined in the code below
2021-07-08 20:51:41 +00:00
