* io.c (io_unread): fix IO#pos with mode 'r' bug on Windows.
If the end of reading buffer is CR, io_unread() needs to unread one
more byte.
[ruby-core:44874] [Bug #6401]
* test/ruby/test_io_m17n.rb (TestIO_M17N#test_pos_with_buffer_end_cr):
add a test for above.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@35613 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (rb_io_eof): use eof() instead of io_fillbuf(). It's because
io_unread() doesn't work properly when reading CRLF with read(length)
and mode 'r'.
[ruby-core:44189][Bug #6271]
* test/ruby/test_io_m17n.rb (TestIO_M17N#test_read_crlf_and_eof):
test for above.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@35331 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (set_binary_mode_with_seek_cur): reorder function qualifiers.
* test/ruby/test_io.rb (TestIO#test_pos_with_getc): added.
see [Bug #6179][ruby-core:43518]
* test/ruby/test_io.rb (TestIO#test_pos_with_getc): updated.
see [ruby-core:43550]
* io.c (static int io_fflush): add the definition.
Use it in set_binary_mode_with_seek_cur().
* io.c (set_binary_mode_with_seek_cur): refactoring to split the
content into io_unread(). Fix the possibility of buffer overflow.
* io.c (io_unread): add new implementation for Windows. Previous one
caused invalid cursor position using IO#pos with OS text mode. New
one fixes the bug.
* test/ruby/test_io_m17n.rb
(TestIO_M17N#test_pos_dont_move_cursor_position): add a test for
above bug.
[ruby-core:43497] [Bug #6179]
* io.c (io_unread): fixed memory leak. report by nagachika via IRC.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@35328 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (io_unread): cast as long the value for extra_max.
[ruby-core:44137] [Bug #6257]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@35256 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* dir.c, file.c, io.c: use rb_sys_fail_path.
* error.c: new functions to deal exceptions with string instances.
* dir.c, file.c, io.c (rb_sys_fail_path): use rb_sys_fail_str.
* test/ruby/test_literal.rb (TestRubyLiteral#test_special_const):
test for https://bugs.php.net/bug.php?id=61095
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@34814 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (Init_IO): use directive hack to make ARGF documentable
in other tools. [ruby-core:42515][Bug #6007]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@34578 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* win32/win32.c, include/ruby/win32.h (rb_w32_fd_is_text): new function.
* win32/win32.c (init_stdhandle): set default mode of stdin as binmode.
* io.c (set_binary_mode_with_seek_cur): new function to replace
SET_BINARY_MODE_WITH_SEEK_CUR macro. now returns previous mode of the
fd and take care of LF in rbuf.
* io.c (do_writeconv): set text mode when needed.
* io.c (io_read): need to change the mode of the IO to binmode
temporally when the length for IO#read, because IO#read with length
must behave so.
* test/ruby/test_io_m17n.rb (TestIO_M17N#est_{read_with_length,
read_with_length_binmode,get[cs]_and_read_with_binmode,
read_with_binmode_and_get[cs],read_write_with_binmode}): tests for
above changes.
all patches are written by Hiroshi Shirosaki. [ruby-core:41496]
[Feature #5714]
* test/ruby/test_io_m17n.rb
(TestIO_M17N#test_{read_with_binmode_and_get[cs]}): only for Windows.
* test/ruby/test_io_m17n.rb (TestIO_M17N#test_{read_with_length,
* io.c (rb_sys_fail_path): move the definition.
Move above for using it in set_binary_mode_with_seek_cur().
* io.c (set_binary_mode_with_seek_cur): fix improper seek cursor.
Seeking file cursor with setting binary mode has possibility to
cause infinite loop. Fixed the bug and refined error handling.
Introduced at r34043.
And cleanups as below.
Remove unnecessary parentheses of `fptr`.
Use return value of setmode().
* test/ruby/test_io_m17n.rb
(TestIO_M17N#test_seek_with_setting_binmode): add a test for abobe.
[ruby-core:41671] [Bug #5714]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@34546 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
------------------------------------------------------------------------
r33662 | usa | 2011-11-08 02:48:11 +0900 (Tue, 08 Nov 2011) | 18 lines
* include/ruby/encoding.h (ECONV_NEWLINE_DECORATOR_READ_MASK,
ECONV_NEWLINE_DECORATOR_WRITE_MASK): new macro.
* io.c (rb_io_extract_modeenc, pipe_open, prep_stdio, argf_next_argv):
set TEXTMODE_NEWLINE_DECORATOR_ON_WRITE for textmode on creating IO
if the flag is available.
* io.c (make_writeconv): drop decorators for reading.
* io.c (make_readconv): drop decorators for writing.
* io.c (do_writeconv): existing writeconv is not the condition to raise
ArgumentError. should check textmode or not.
* test/ruby/test_io_m17n.rb
(TestIO_M17N#test_{cr,lf,crlf}_decorator_on_stdout): test above
changes.
------------------------------------------------------------------------
* include/ruby/encoding.h (ECONV_NEWLINE_DECORATOR_READ_MASK,
ECONV_NEWLINE_DECORATOR_WRITE_MASK): new macro.
* io.c (rb_io_extract_modeenc, pipe_open, prep_stdio, argf_next_argv):
set TEXTMODE_NEWLINE_DECORATOR_ON_WRITE for textmode on creating IO
if the flag is available.
* io.c (make_writeconv): drop decorators for reading.
* io.c (make_readconv): drop decorators for writing.
* io.c (do_writeconv): existing writeconv is not the condition to raise
ArgumentError. should check textmode or not.
* test/ruby/test_io_m17n.rb
(TestIO_M17N#test_{cr,lf,crlf}_decorator_on_stdout): test above
changes.
* io.c (argf_next_argv): wrong timing of setting ecflags.
fixed the failure of TestArgf#test_textmode introduced at r33662.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@34542 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (rb_io_fsync,rb_io_fdatasync): release GVL during fsync().
fsync() and fdatasync() may take a long time on slow disks and/or
if there is much dirty data.
Patch by Eric Wong. [Feature #5665] [ruby-core:41247]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@34451 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (io_fwrite): call rb_w32_write_console() only if FMODE_TTY is
set. this is the one of the reason of IO writing slowness of Windows
in 1.9.3 or later.
[Backport #5592]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@34436 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
so, if the io is text mode and has ECONV_UNIVERSAL_NEWLINE_DECORATOR
flag, use crlf_newline converter for writing.
this change fixes the problem about the luck of CR up Kernel.p and
Kernel.puts to stdout/stderr on Windows.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@33660 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (rb_io_flush, rb_io_rewind): fsync() here.
These pathces are backports of trunk r33651 for [Bug #5585]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@33658 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* io.c (argf_next_argv): open in default text mode.
[ruby-core:39234] [Bug #5268]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_3@33211 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
RB_RESERVED_FD_P): added. This C API is to limit to access
fds which are used by RubyVM internal. In this version of
CRuby, return 1 if fd is communication pipe.
If your application needs to close all file descriptors to
preent resource leak, skip internal fds using this C API.
We also define a macro RB_RESERVED_FD_P(fd). So you can write
#ifndef RB_RESERVED_FD_P
#define RB_RESERVED_FD_P(fd) 0
#endif
for Ruby 1.9.2 or previous version to write compatible extensions.
See [ruby-core:37727]
* thread_win32.c (rb_reserved_fd_p): added (return 0 for any fds).
* io.c (rb_io_initialize): raise ArgumentError if given fd is reserved by Ruby.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@32394 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
been cleared by another thread while io_fillbuf() is waiting at
select(). a patch in [ruby-core:37197] by Hiroshi Shirosaki
<h.shirosaki AT gmail.com>. fixed#3840
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@32169 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Fixed by Eric Wong. [ruby-core:36150]
(maygvl_copy_stream_wait_readwrite): removed.
(nogvl_copy_stream_sendfile): use nogvl_copy_stream_wait_write and
maygvl_copy_stream_wait_read instead of
maygvl_copy_stream_wait_readwrite.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@32019 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
based on the patch by Eric Wong. [ruby-core:36156]
* vm_core.h (rb_thread_call_with_gvl): don't declare here.
* thread.c: include internal.h.
(rb_thread_execute_interrupts): new function.
* internal.h (rb_thread_execute_interrupts): declared.
(rb_thread_call_with_gvl): declared.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@31971 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* test/ruby/test_io.rb: Test for File.write, File.binwrite.
* NEWS: News for above.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@31902 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
DEFAULT_TEXTMODE in call of io_set_encoding(), and of write IO is
also set as it in call of io_new_instance() via rb_protect().
so, if DEFAULT_TEXTMODE is not 0, we should check the result of
extract_binmode() and avoid crush of default IO mode and the result.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@31859 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
because boron chkbuild test result says, An old linux kernel ignore O_CLOEXEC
silently instead of return an error. It may lead to bring new security risk.
So, we have to be pending it until finish to implement proper fallback logic.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@31643 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
