The new event processing architecture changed `jobwait()` semantics: Only one
job is processed at time since process_wait only focuses on one queue.
This fixes the problem with a few changes:
- Allow the event queue polled by `process_wait` to be overriden by a new
argument.
- Allow the parent queue to be overriden with `queue_replace_parent`
- Create a temporary queue that serves as the parent for all jobs passed to
`jobwait()`
Since pty events are queued, it is possible that the reads will be reordered.
Example scenario:
In the terminal you have output combined from stdout and stderr. A program
generates output, first you have some output on stdout, then output on stderr,
output on stdout, output on stderr,... The whole output should be interleaved
from both streams.
Each output generates a read_event and they are placed in the same queue. If the
queue is processed, the first read_event will send the whole stdout output to
the terminal (on_job_output() consumes the whole buffer). The next read_event is
similar for stderr. The remaining read events do nothing because now both
RBuffer are already empty. So the terminal would show first the stdout output
and after that the stderr output.
This commit fixes the problem by disabling stderr stream in pty processes.
That's ok because they all represent the same stream(duplicate file
descriptors), plus one stream is simpler to deal with.
- Improve the implementation of deferred/immediate events.
- Use the new queue module to change how/when events are queued/processed by
giving a private queue to each emitter.
- Immediate events(which only exist to break uv_run recursion) are now
represented in the `loop->fast_events` queue.
- Events pushed to child queues are propagated to the event loop main queue and
processed as K_EVENT keys.
Compiler warning was found in [QuickBuild logs][1] from [this page][2]. GCC and
clang on travis appear to be fine. Relevant log parts:
Step Log (master>buildall>build-node?testNode=linux-64>build-and-run-tests>build-and-run-tests-parameterized?buildType=Release>configure-neovim-and-build-nvim)
<…>
16:26:31,364 WARN - /home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/eval.c: In function ‘f_msgpackdump’:
16:26:31,364 WARN - /home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/eval.c:12371:26: error: ‘cur_tv’ may be used uninitialized in this function [-Werror=uninitialized]
16:26:31,364 WARN - /home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/eval.c:12328:21: note: ‘cur_tv’ was declared here
16:26:31,938 WARN - cc1: all warnings being treated as errors
[1]: http://neovim-qb.szakmeister.net/wicket/page?4-1.ILinkListener-content-buildTab-panel-errorContainer-steps-5-logLink
[2]: http://neovim-qb.szakmeister.net/build/2099/overview
glibc on i386 seems to have an optimization that makes it harder to call
some math functions indirectly from inside an inlined function, causing
compile-time errors with some versions of gcc. This removes inlining on
that platform.
Closes#3071
In Vim, which doesn't true colors, synIDattr('Foo', 'fg') returns either
ctermfg or guifg depending on whether vim or gvim is running.
True colors naturally use GUI colors, so synIDattr() has to be adapted to
return guifg, if a TUI with enabled true colors is used.
The same error checking/recovery code was duplicated among the f_acos,
f_asin, f_atan, f_ceil, f_cos, f_cosh, f_exp, f_floor, f_log, f_log10,
f_round, f_sin, f_sinh, f_sqrt, f_tan, f_tanh and f_trunc functions.
This commit moves that code into a wrapper function.
`trunc` is not in C90, but it is in C99, which is what neovim targets,
so we use it here.
- New libuv/pty process abstraction with simplified API and no globals.
- Remove nvim/os/job*. Jobs are now a concept that apply only to programs
spawned by vimscript job* functions.
- Refactor shell.c/channel.c to use the new module, which brings a number of
advantages:
- Simplified API, less code
- No slots in the user job table are used
- Not possible to acidentally receive data from vimscript
- Implement job table in eval.c, which is now a hash table with unilimited job
slots and unique job ids.
- Simplify RStream/WStream API and make it more consistent with libuv.
- Move into the event loop layer(event subdirectory)
- Remove uv_helpers module.
- Simplify job/process internal modules/API.
- Unify RStream and WStream into a single structure. This is necessary because
libuv streams can be readable and writable at the same time(and because the
uv_helpers.c hack to associate multiple streams with libuv handle was removed)
- Make struct definition public, allowing more flexible/simple memory
management by users of the module.
- Adapt channel/job modules to cope with the changes.
- Add event loop abstraction module under src/nvim/event. The
src/nvim/event/loop module replaces src/nvim/os/event
- Remove direct dependency on libuv signal/timer API and use the new abstraction
instead.
- Replace all references to uv_default_loop() by &loop.uv, a new global variable
that wraps libuv main event loop but allows the event loop functions to be
reused in other contexts.
Extract the RBuffer class from rstream.c and reimplement it as a ring buffer,
a more efficient version that doesn't need to relocate memory.
The old rbuffer_read/rbuffer_write interfaces are kept for simple
reading/writing, and the RBUFFER_UNTIL_{FULL,EMPTY} macros are introduced to
hide wrapping logic when more control is required(such as passing the buffer
pointer to a library function that writes directly to the pointer)
Also add a basic infrastructure for writing helper C files that are only
compiled in the unit test library, and use this to write unit tests for RBuffer
which contains some macros that can't be accessed directly by luajit.
Helped-by: oni-link <knil.ino@gmail.com>
Reviewed-by: oni-link <knil.ino@gmail.com>
Reviewed-by: Scott Prager <splinterofchaos@gmail.com>
Reviewed-by: Justin M. Keyes <justinkz@gmail.com>
Reviewed-by: Michael Reed <m.reed@mykolab.com>
suspicious_sizeof: Passing argument 168UL /* sizeof (ufunc_T) */ * todo
to function xmalloc and then casting the return value to ufunc_T ** is
suspicious.
For any of these functions, if {cmd} is a string, execute
"&shell &shellcmdflag '{cmd}'", or simply {cmd} if it's a list.
In termopen(), if the 'name' option is not supplied, try to guess using
'{cmd}' (string) or {cmd}[0] (list). Simplify ex_terminal to use the
string form of termopen().
termopen: get name from argument
Convert list_to_argv to tv_to_argv.
Helped-by: Björn Linse <@bfredl>
Helped-by: oni-link <knil.ino@gmail.com>
Helped-by: Thiago de Arruda <@tarruda>
Old behaviour: termopen('cmd') would run `&shell &shcf "cmd"`, which
caused the functional tests to fail on some systems due to the process
not "owning" the terminal. Also, it is inconsistent with jobstart().
Modify termopen() so that &shell is not invoked, but maintain the old
behaviour with :terminal. Factor the common code for building the
argument vector from jobstart() and modify the functional tests to call
termopen() instead of :terminal (fixes#2354).
Also:
* Add a 'name' option for termopen() so that `:terminal {cmd}` produces
a buffer named "term//{cwd}/{cmd}" and termopen() users can customize
the name.
* Update the documentation.
* Add functional tests for `:terminal` sinse its behaviour now differs
from termopen(). Add "test/functional/fixtures/shell-test.c" and move
"test/functional/job/tty-test.c" there, too.
Helped-by: Justin M. Keyes <@justinmk>
