kcar RubyGem user+dev discussion/patches/pulls/bugs/help
 help / color / mirror / code / Atom feed
Search results ordered by [date|relevance]  view[summary|nested|Atom feed]
thread overview below | download mbox.gz: |
* [ANN] kcar 0.7.0 - bytestream to Rack response converter
@ 2020-02-21  2:32  5% Eric Wong
  0 siblings, 0 replies; 4+ results
From: Eric Wong @ 2020-02-21  2:32 UTC (permalink / raw)
  To: kcar-public

kcar features an HTTP parser that will convert a bytestream into a
3-element array suitable for use as a Rack response.  It is IO interface
agnostic, so it may be used with HTTP streams over Unix domain sockets,
regular files, FIFOs, StringIOs as well as traditional TCP sockets.

* homepage: https://yhbt.net/kcar/
* public inbox: kcar-public@yhbt.net
* git clone https://yhbt.net/kcar.git
* mailing list archives: https://yhbt.net/kcar-public/

Changes:

46 changes since 0.6.0 (2015-08-04):
      README: fix reference to HTTP git viewer
      doc: move site to HTTPS
      pkg.mk: avoid network for "gem install"
      gemspec: use SPDX compatible terms for the license(s)
      archive/slrnpull.conf: add a note explaining the purpose
      drop rb_str_set_len compatibility replacement
      remove rb_str_modify workaround
      TypedData C-API conversion
      test_parser: add lone CR test
      reduce parser size to 88 bytes on 64-bit
      extconf: remove unneeded -fPIC CFLAGS
      rely on String#-@ (str_uminus) to dedupe headers
      update comment about freezing values
      http: reject non-LWS CTL chars (0..31 + 127) in field values
      doc: remove references to a server
      update documentation for contributions, remove private address
      doc: minor updates to describe classes, better
      response: remove unnecessary constant alias
      response: more documentation cleanups
      favor require_relative to speed up loading
      response: remove Ruby 1.8-compatibility check
      fix signedness check on 32-bit systems
      shorten and improve readability of assertion
      HACKING: remove copy+pasted line about N
      olddoc: include NNTP archive link
      gemspec: remove olddoc dev dependency
      README: add info about mailing list subscription
      nodoc Kcar::VERSION
      pkg.mk: support VALGRIND variable for unit tests
      introduce new str_new_dd_freeze internal function
      begin implementing request parsing
      favor bitfields instead flags + macros
      implement request parsing with tests
      pkg.mk: enable warnings by default for tests
      filter_body: rename variables to be like memcpy(3)
      flesh out filter_body for request parsing
      do not assume SERVER_PORT
      do not set "HTTP/0.9" for pre-1.0 requests
      always set non-negative Content-Length for requests
      avoid String#-@ call on request parsing under Ruby 2.6
      request: set env["FRAGMENT"] for WebDAV litmus test
      extconf: fix rb_hash_aset deduplication test
      use rb_gc_register_mark_object
      website: use dark216 to save electricity
      doc: update URLs to point to YHBT.net
      doc: update git:// URLs to HTTPS

^ permalink raw reply	[relevance 5%]

* [PATCH 01/11] introduce new str_new_dd_freeze internal function
  @ 2018-12-01 13:31  6% ` Eric Wong
  0 siblings, 0 replies; 4+ results
From: Eric Wong @ 2018-12-01 13:31 UTC (permalink / raw)
  To: kcar-public

This seems like it will be a common pattern to create
and immediately dedupe a string.  In the future, we may
be able to save some memory by eliding temporary object
creation or releasing heap memory via rb_str_resize(..., 0)
---
 ext/kcar/extconf.rb | 15 ++++++++++-----
 ext/kcar/kcar.rl    |  9 +++++++--
 2 files changed, 17 insertions(+), 7 deletions(-)

diff --git a/ext/kcar/extconf.rb b/ext/kcar/extconf.rb
index 89ac586..b65846a 100644
--- a/ext/kcar/extconf.rb
+++ b/ext/kcar/extconf.rb
@@ -6,15 +6,20 @@ dir_config("kcar")
 have_macro("SIZEOF_OFF_T", "ruby.h") or check_sizeof("off_t", "sys/types.h")
 have_macro("SIZEOF_LONG", "ruby.h") or check_sizeof("long", "sys/types.h")
 
-uminus_dedupe = false
+message('checking if String#-@ (str_uminus) dedupes... ')
 begin
-  # oddly, opt_str_freeze is not always effective:
-  # https://bugs.ruby-lang.org/issues/13282
   a = -(%w(t e s t).join)
   b = -(%w(t e s t).join)
-  uminus_dedupe = a.object_id == b.object_id
+  if a.equal?(b)
+    $CPPFLAGS += " -DSTR_UMINUS_DEDUPE=1 "
+    message("yes\n")
+  else
+    $CPPFLAGS += " -DSTR_UMINUS_DEDUPE=0 "
+    message("no, needs Ruby 2.5+\n")
+  end
 rescue NoMethodError
+  $CPPFLAGS += " -DSTR_UMINUS_DEDUPE=0 "
+  message("no, String#-@ not available\n")
 end
-$CFLAGS += " -DSTR_UMINUS_DEDUPE=#{uminus_dedupe ? 1 : 0}"
 
 create_makefile("kcar_ext")
diff --git a/ext/kcar/kcar.rl b/ext/kcar/kcar.rl
index cbcfa97..79f65db 100644
--- a/ext/kcar/kcar.rl
+++ b/ext/kcar/kcar.rl
@@ -106,6 +106,11 @@ static VALUE str_dd_freeze(VALUE str)
   return str;
 }
 
+static VALUE str_new_dd_freeze(const char *ptr, long len)
+{
+  return str_dd_freeze(rb_str_new(ptr, len));
+}
+
 static VALUE stripped_str_new(const char *str, long len)
 {
   long end;
@@ -161,7 +166,7 @@ status_phrase(struct http_parser *hp, VALUE hdr, const char *ptr, size_t len)
 {
   long nr;
 
-  hp->status = str_dd_freeze(rb_str_new(ptr, len));
+  hp->status = str_new_dd_freeze(ptr, len);
 
   /* RSTRING_PTR is null terminated, ptr is not */
   nr = strtol(RSTRING_PTR(hp->status), NULL, 10);
@@ -238,7 +243,7 @@ static void write_value(VALUE hdr, struct http_parser *hp,
   vlen = LEN(mark, p);
   VALIDATE_MAX_LENGTH(vlen, FIELD_VALUE);
   VALIDATE_MAX_LENGTH(flen, FIELD_NAME);
-  f = str_dd_freeze(rb_str_new(fptr, (long)flen));
+  f = str_new_dd_freeze(fptr, (long)flen);
   v = stripped_str_new(vptr, (long)vlen);
 
   /* needs more tests for error-checking here */

^ permalink raw reply related	[relevance 6%]

* [PATCH 1/7] introduce new str_new_dd_freeze internal function
  2017-04-19 22:30  6% [PATCH 0/7] request parsing bits Eric Wong
@ 2017-04-19 22:30  7% ` Eric Wong
  0 siblings, 0 replies; 4+ results
From: Eric Wong @ 2017-04-19 22:30 UTC (permalink / raw)
  To: kcar-public

This seems like it will be a common pattern to create
and immediately dedupe a string.  In the future, we may
be able to save some memory by eliding temporary object
creation or releasing heap memory via rb_str_resize(..., 0)
---
 ext/kcar/kcar.rl | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/ext/kcar/kcar.rl b/ext/kcar/kcar.rl
index cbcfa97..79f65db 100644
--- a/ext/kcar/kcar.rl
+++ b/ext/kcar/kcar.rl
@@ -106,6 +106,11 @@ static VALUE str_dd_freeze(VALUE str)
   return str;
 }
 
+static VALUE str_new_dd_freeze(const char *ptr, long len)
+{
+  return str_dd_freeze(rb_str_new(ptr, len));
+}
+
 static VALUE stripped_str_new(const char *str, long len)
 {
   long end;
@@ -161,7 +166,7 @@ status_phrase(struct http_parser *hp, VALUE hdr, const char *ptr, size_t len)
 {
   long nr;
 
-  hp->status = str_dd_freeze(rb_str_new(ptr, len));
+  hp->status = str_new_dd_freeze(ptr, len);
 
   /* RSTRING_PTR is null terminated, ptr is not */
   nr = strtol(RSTRING_PTR(hp->status), NULL, 10);
@@ -238,7 +243,7 @@ static void write_value(VALUE hdr, struct http_parser *hp,
   vlen = LEN(mark, p);
   VALIDATE_MAX_LENGTH(vlen, FIELD_VALUE);
   VALIDATE_MAX_LENGTH(flen, FIELD_NAME);
-  f = str_dd_freeze(rb_str_new(fptr, (long)flen));
+  f = str_new_dd_freeze(fptr, (long)flen);
   v = stripped_str_new(vptr, (long)vlen);
 
   /* needs more tests for error-checking here */
-- 
EW


^ permalink raw reply related	[relevance 7%]

* [PATCH 0/7] request parsing bits
@ 2017-04-19 22:30  6% Eric Wong
  2017-04-19 22:30  7% ` [PATCH 1/7] introduce new str_new_dd_freeze internal function Eric Wong
  0 siblings, 1 reply; 4+ results
From: Eric Wong @ 2017-04-19 22:30 UTC (permalink / raw)
  To: kcar-public

There can be more optimizations done(*), but I might push
out a prerelease in a bit and start running it on on real
code for a bit.

7 patches:
      introduce new str_new_dd_freeze internal function
      begin implementing request parsing
      favor bitfields instead flags + macros
      implement request parsing with tests
      pkg.mk: enable warnings by default for tests
      filter_body: rename variables to be like memcpy(3)
      flesh out filter_body for request parsing

 ext/kcar/kcar.rl             |  583 +++++++++++++++++---
 ext/kcar/kcar_http_common.rl |   36 +-
 pkg.mk                       |    2 +-
 test/test_parser.rb          |    1 +
 test/test_request_parser.rb  | 1219 ++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 1748 insertions(+), 93 deletions(-)


(*) and they'll be ported to that other-parser-that-shall-not-be-named

^ permalink raw reply	[relevance 6%]

Results 1-4 of 4 | reverse | options above
-- pct% links below jump to the message on this page, permalinks otherwise --
2017-04-19 22:30  6% [PATCH 0/7] request parsing bits Eric Wong
2017-04-19 22:30  7% ` [PATCH 1/7] introduce new str_new_dd_freeze internal function Eric Wong
2018-12-01 13:31     [PATCH v2] request parsing bits Eric Wong
2018-12-01 13:31  6% ` [PATCH 01/11] introduce new str_new_dd_freeze internal function Eric Wong
2020-02-21  2:32  5% [ANN] kcar 0.7.0 - bytestream to Rack response converter Eric Wong

Code repositories for project(s) associated with this public inbox

	https://yhbt.net/kcar.git/

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).