| Date | Commit message (Collapse) |
|---|
|
Non-socket activation users will want to install Net::Server
for daemonization, pid file writing, and user/group switching.
|
|
We handle encoding-related things elsewhere.
|
|
No need to create a new sub which kill ourselves $$ when we can
invoke worker_quit directly.
|
|
Not that these subs are repeatedly created, but this makes
the code easier-to-review and these callbacks are idempotent
anyways.
|
|
We do not want to be accepting connections during graceful
shutdown because another new process is likely taking over.
This also allows us to free up the listener case another
(independent) process wants to claim it.
|
|
We do not need line buffering, here; so favor sysread to
bypass extra copies which may be done by normal read.
|
|
IO::Handle->new_from_fd has existed since at least 1996,
so it should be safe to depend on at this point.
|
|
Listening on Unix domain sockets can be convenient for running
behind reverse proxies, avoiding port conflicts, limiting access,
or avoiding the overhead (if any) of TCP over loopback.
|
|
This allows us to share more code between daemons and avoids
having to make additional syscalls for preparing REMOTE_HOST
and REMOTE_PORT in the PSGI env in -httpd.
This will also make supporting HTTP (and NNTP) over Unix sockets
easier in a future commit.
|
|
We've distilled the daemon code into one public function ("run"),
so avoid polluting the main namespace and just have users
prefix with the full package name for this rarely-used class.
|
|
Ugh, this enabled-iff-xapian-is-available code really
needs better testing...
|
|
Message-IDs should not be MIME encoded, but in case they are,
use the raw form for compatibility with ssoma and possibly
other tools. This prevents a potential problem where a
malicious client could confuse our storage layer into indexing
incorrect contents.
|
|
Better to throw the error back to the client ASAP if we're
out-of-descriptors. We will need to implement idle client
expiration for long-lived HTTP connections.
|
|
We'll have to use it some more before deciding it is a public
interface. I do hope for it to be a usable public interface
one day for other users.
|
|
We need to ensure close on handles tied to this class
get the same errors a normal "close" in Perl gets.
|
|
It seems common for users to end statements with URLs,
while it is rare for a URL itself to end with a '.' or ';'.
So make a guess and assume the URL was intended to not
include the trailing '.' or ';'
|
|
This will allow us to more easily reuse it elsewhere.
|
|
We now keep intermediate blank lines in messages, since it
could be used to denote logical gaps in the message
(such as giving readers a chance to opt out of "spoiler"
information).
However leading blank lines, trailing blank lines, and
trailing whitespace have no useful value we can discern;
so drop those entirely to prevent clients from eating up
vertical whitespace.
|
|
It's often not that much information and may be useful
to reduce HTTP requests a reader will want to make.
|
|
We do not need to load Plack::Request outside of WWW anymore.
|
|
Apache2 mod_perl does not give us a real file handle, so
we must translate that before giving that to git-http-backend(1).
Also, parse the Status: correctly for errors since we failed to
set %ENV properly before the previous fix for SpawnPP
|
|
We cannot modify %ENV directly under mod_perl (even after forking!),
so use env(1) instead to pass the environment.
|
|
It is not needed as we know git uses CRLF termination.
|
|
This makes for better compile-time checking and also helps
document which calls are private for HTTP and NNTP.
While we're at it, use IO::Handle::* functions procedurally,
too, since we know we're working with native glob handles.
|
|
For error messages intended to show user error (e.g. giving
invalid options), we add a newline ("\n") at the end to
polluting the output with location information.
However, for diagnosing non-user-triggered errors, we should
show the location of where the error occured.
|
|
Checking the time is nearly free on modern systems with
vDSO/vsyscall/similar while sprintf is always expensive.
|
|
It may not be obvious where we are when we enter the event_write
callback. Hopefully this clarifies things.
|
|
Just in case we screwed up somewhere, we need to match up
syswrite to sysseek and we also favor procedural calls for
native types.
|
|
Perl may complain about exit not being executed, but not die.
|
|
Oops :x
|
|
HTTP responses may be long-running or requests may be slow or
pipelined. Ensure we don't kill them off prematurely.
|
|
We can rely on timely auto-destruction based on reference
counting; reducing the chance of redundant close(2) calls
which may hit the wront FD.
We do care about certain close calls (e.g. writing to a buffered
IO handle) if we require error-checking for write-integrity. In
other cases, let things go out-of-scope so it can be freed
automatically after use.
|
|
While empty or "0" should never appear, this allows the
reviewer to think and know less about the context in which
this check is done.
|
|
Not sure how, but this should've always been AGPL-3.0+ like
the rest of the code, not GPL-3.0+
|
|
This is necessary since we want to be able to do arbitrary redirects
via the popen interface. Oh well, we'll be a little slower for now
for users without vfork. vfork users will get all the performance
benefits.
|
|
Oops :x
|
|
This means we always load the PSGI server code early for
-httpd. This may make things less compatible with existing
PSGI/Plack apps, but we prioritize our httpd for the uses
of public-inbox itself, first.
And any existing PSGI/Plack app which wants to may adapt
themselves to being preload-friendly.
|
|
We must stash the error correctly when nesting evals, oops :x
|
|
This should reduce overhead of spawning git processes
from our long-running httpd and nntpd servers.
|
|
Under Linux, vfork maintains constant performance as
parent process size increases. fork needs to prepare pages
for copy-on-write, requiring a linear scan of the address
space.
|
|
No point in comparing an empty string; length() is only
potentially expensive on big strings.
|
|
Writing a read-only IMAP server isn't out-of-scope, either,
but I've never studiied the IMAP protocol, much, unlike HTTP/1.x
or even NNTP.
|
|
We want to preload as much as possible in -httpd when forking
to save memory via CoW.
|
|
Some linkifiers to create invalid HTTP links when it sees a
link intended for NNTP services. This means we may see links
to news.public-inbox.org/inbox.comp.mail.public-inbox.meta
point to "http://" on port 80 instead of 119. Try to
redirect users to http://public-inbox.org/meta/ in this case.
|
|
All URL generation in dynamic HTTP pages should be capable of
generating "https" or "http" URLs depending on the user's
preference.
|
|
This will allow us to more easily read and test later.
|
|
We cannot modify elements in any shared data strucutures
shared between requests. Oops!
|
|
We will be falling back and cascading to newsgroup lookups, later.
|
|
Danga::Socket will die on us if we hit the base implementations.
|
|
Even with output buffering disabled via IO::Handle::autoflush,
writes are not atomic unless it is a single argument passed to
"print". Multiple arguments to "print" will show up as multiple
calls to write(2) instead of a single, atomic writev(2).
|