]> Sergey Matveev's repositories - public-inbox.git/log
public-inbox.git
4 years agoci: require IO::KQueue on FreeBSD, for now
Eric Wong [Mon, 24 Jun 2019 02:52:51 +0000 (02:52 +0000)]
ci: require IO::KQueue on FreeBSD, for now

We'll likely replace IO::KQueue (at least on FreeBSD) using
a pure-Perl syscall()-based version since syscall numbers are
consistent across architectures on FreeBSD and easy to maintain.

IO::KQueue->EV_SET is also shockingly inefficient in that it
calls kqueue() as much as epoll_ctl.

4 years agonntp: lazily allocate and stash rbuf
Eric Wong [Mon, 24 Jun 2019 02:52:50 +0000 (02:52 +0000)]
nntp: lazily allocate and stash rbuf

Allocating a per-client buffer up front is unnecessary and
wastes a hash slot.  For the majority of (non-malicious)
clients, we won't need to store rbuf in a long-lived object
associated with a client socket at all.

This saves around 10M on 64-bit with 20K connected-but-idle
clients.

4 years agods: flush_write runs ->write callbacks even if closed
Eric Wong [Mon, 24 Jun 2019 02:52:49 +0000 (02:52 +0000)]
ds: flush_write runs ->write callbacks even if closed

We may need to rely on cleanup code running in enqueued
callbacks, so ensure we call it when flush_write happens.

4 years agonntp: simplify long response logic and fix nesting
Eric Wong [Mon, 24 Jun 2019 02:52:48 +0000 (02:52 +0000)]
nntp: simplify long response logic and fix nesting

We can get rid of the {long_res} field and reuse the write
buffer ordering logic to prevent nesting of responses from
requeue.

On FreeBSD, this fixes a problem of callbacks firing twice
because kqueue as event_step is now our only callback entry
point.

There's a slight change in the stdout "logging" format, in
that we can no longer distinguish between writes blocked
due to slow clients or deferred long responses.  Not sure
if this affects anybody parsing logs or not, but preserving
the old format could prove expensive and not worth the
effort.

4 years agods: always use EV_ADD with EV_SET
Eric Wong [Mon, 24 Jun 2019 02:52:47 +0000 (02:52 +0000)]
ds: always use EV_ADD with EV_SET

kqueue EV_ONESHOT semantics are different than epoll
EPOLLONESHOT.  epoll only disables watches for that event while
keeping the item in the rbtree for future EPOLL_CTL_MOD.  kqueue
removes the watch from the filter set entirely, necessitating
the use of EV_ADD for future modifications.

4 years agonntp: reduce allocations for greeting
Eric Wong [Mon, 24 Jun 2019 02:52:46 +0000 (02:52 +0000)]
nntp: reduce allocations for greeting

No need to allocate a new PerlIO::scalar filehandle for every
client, instead we can now pass the same CODE reference which
calls DS->write on a reused string reference.

4 years agods: allow ->write callbacks to syswrite directly
Eric Wong [Mon, 24 Jun 2019 02:52:45 +0000 (02:52 +0000)]
ds: allow ->write callbacks to syswrite directly

We can bypass buffering when wbuf is empty when it's called
from a CODE reference passed to ->write.

4 years agodaemon: use SSL_MODE_RELEASE_BUFFERS
Eric Wong [Mon, 24 Jun 2019 02:52:44 +0000 (02:52 +0000)]
daemon: use SSL_MODE_RELEASE_BUFFERS

34K per idle connection adds up to large amounts of memory;
especially with the speed of malloc nowadays compared to the
cost of cache misses or worse, swapping.

4 years agot/nntpd-tls: slow client connection test
Eric Wong [Mon, 24 Jun 2019 02:52:43 +0000 (02:52 +0000)]
t/nntpd-tls: slow client connection test

We need to ensure slowly negotiating TLS clients don't block
the event loop.  This is why I added the size check of
{wbuf} before and after calling the CODE ref in DS::flush_write.

4 years agonntp: call SSL_shutdown in normal cases
Eric Wong [Mon, 24 Jun 2019 02:52:42 +0000 (02:52 +0000)]
nntp: call SSL_shutdown in normal cases

This is in accordance with TLS standards and will be needed
to support session caching/reuse in the future.  However, we
don't issue shutdown(2) since we know not to inadvertantly
share our sockets with other processes.

4 years agods|nntp: use CORE::close on socket
Eric Wong [Mon, 24 Jun 2019 02:52:41 +0000 (02:52 +0000)]
ds|nntp: use CORE::close on socket

IO::Socket::SSL will try to re-bless back to the original class
on TLS negotiation failure.  Unfortunately, the original class
is 'GLOB', and re-blessing to 'GLOB' takes away all the IO::Handle
methods, because Filehandle/IO are a special case in Perl5.
Anyways, since we already use syswrite() and sysread() as functions
on our socket, we might as well use CORE::close(), as well (and
it plays nicely with tied classes).

4 years agodaemon: map inherited sockets to well-known schemes
Eric Wong [Mon, 24 Jun 2019 02:52:40 +0000 (02:52 +0000)]
daemon: map inherited sockets to well-known schemes

I don't want to specify "--listen" in my systemd .service files,
so map 563 to NNTPS automatically (and 443 to HTTPS, but HTTPS
support doesn't work, yet).

4 years agocerts/create-certs.perl: fix cert validity on 32-bit
Eric Wong [Mon, 24 Jun 2019 02:52:39 +0000 (02:52 +0000)]
certs/create-certs.perl: fix cert validity on 32-bit

If I'm still alive, I won't be coding after 2038 :<

4 years agonntp: NNTPS and NNTP+STARTTLS working
Eric Wong [Mon, 24 Jun 2019 02:52:38 +0000 (02:52 +0000)]
nntp: NNTPS and NNTP+STARTTLS working

It kinda, barely works, and I'm most happy I got it working
without any modifications to the main NNTP::event_step callback
thanks to the DS->write(CODE) support we inherited from
Danga::Socket.

4 years agonntp: wait for writability before sending greeting
Eric Wong [Mon, 24 Jun 2019 02:52:37 +0000 (02:52 +0000)]
nntp: wait for writability before sending greeting

This will be needed for NNTPS support, since we need
to negotiate the TLS connection before writing the
greeting and we can reuse the existing buffer layer
to enqueue writes.

4 years agods: deal better with FS-related errors IO buffers
Eric Wong [Mon, 24 Jun 2019 02:52:36 +0000 (02:52 +0000)]
ds: deal better with FS-related errors IO buffers

Instead of ENOMEM (or fragmentation/swap storms), using tempfile
buffers opens us up to filesystem and storage-related errors
(e.g. ENOSPC, EFBIG, EIO, EROFS).  Log these errors, drop the
particular client, and try to limp by with whateve we have left.

4 years agoallow use of PerlIO layers for filesystem writes
Eric Wong [Mon, 24 Jun 2019 02:52:35 +0000 (02:52 +0000)]
allow use of PerlIO layers for filesystem writes

It may make sense to use PerlIO::mmap or PerlIO::scalar for
DS write buffering with IO::Socket::SSL or similar (since we can't
use MSG_MORE), so that means we need to go through buffering
in userspace for the common case; while still being easily
compatible with slow clients.

And it also simplifies GitHTTPBackend slightly.

Maybe it can make sense for HTTP input buffering, too...

4 years agonntp: simplify re-arming/requeue logic
Eric Wong [Mon, 24 Jun 2019 02:52:34 +0000 (02:52 +0000)]
nntp: simplify re-arming/requeue logic

We can be smarter about requeuing clients to run and avoid
excessive epoll_ctl calls since we can trust event_step to do
the right thing depending on the state of the client.

4 years agods: hoist out do_read from NNTP and HTTP
Eric Wong [Mon, 24 Jun 2019 02:52:33 +0000 (02:52 +0000)]
ds: hoist out do_read from NNTP and HTTP

Both NNTP and HTTP have common needs and we can factor
out some common code to make dealing with IO::Socket::SSL
easier.

4 years agohttp|nntp: be explicit about bytes::length on rbuf
Eric Wong [Mon, 24 Jun 2019 02:52:32 +0000 (02:52 +0000)]
http|nntp: be explicit about bytes::length on rbuf

It should not matter because our rbuf is always from
a socket without encoding layers, but this makes things
easier to follow.

4 years agods: remove pointless exit calls
Eric Wong [Mon, 24 Jun 2019 02:52:31 +0000 (02:52 +0000)]
ds: remove pointless exit calls

They're never called; the only way to break out of that loop
is the PostEventLoop callback.

4 years agoevcleanup: replace _run_asap with `event_step' callback
Eric Wong [Mon, 24 Jun 2019 02:52:30 +0000 (02:52 +0000)]
evcleanup: replace _run_asap with `event_step' callback

No point in keeping a one-line wrapper sub around.

4 years agods: pass $self to code references
Eric Wong [Mon, 24 Jun 2019 02:52:29 +0000 (02:52 +0000)]
ds: pass $self to code references

We can reduce the amount of short-lived anonymous subs we
create by passing $self to code references.

4 years agohttp: don't pass extra args to PublicInbox::DS::close
Eric Wong [Mon, 24 Jun 2019 02:52:28 +0000 (02:52 +0000)]
http: don't pass extra args to PublicInbox::DS::close

YAGNI

Followup-to: commit 30ab5cf82b9d47242640f748a0f9a088ca783e32
("ds: reduce Errno imports and drop ->close reason")

4 years agods: favor `delete' over assigning fields to `undef'
Eric Wong [Mon, 24 Jun 2019 02:52:27 +0000 (02:52 +0000)]
ds: favor `delete' over assigning fields to `undef'

This is cleaner in most cases and may allow Perl to reuse memory
from unused fields.

We can do this now that we no longer support Perl 5.8; since
Danga::Socket was written with struct-like pseudo-hash support
in mind, and Perl 5.9+ dropped support for pseudo-hashes over
a decade ago.

4 years agohttp|nntp: favor "$! == EFOO" over $!{EFOO} checks
Eric Wong [Mon, 24 Jun 2019 02:52:26 +0000 (02:52 +0000)]
http|nntp: favor "$! == EFOO" over $!{EFOO} checks

Integer comparisions of "$!" are faster than hash lookups.

See commit 6fa2b29fcd0477d126ebb7db7f97b334f74bbcbc
("ds: cleanup Errno imports and favor constant comparisons")
for benchmarks.

4 years agoqspawn: describe where `$rpipe' come from
Eric Wong [Mon, 24 Jun 2019 02:52:25 +0000 (02:52 +0000)]
qspawn: describe where `$rpipe' come from

It wasn't immediately obvious to me after several months of
not looking at this code.

4 years agospawn: remove `Blocking' flag handling
Eric Wong [Mon, 24 Jun 2019 02:52:24 +0000 (02:52 +0000)]
spawn: remove `Blocking' flag handling

Instead, the O_NONBLOCK flag is set by PublicInbox::HTTPD::Async;
and we won't be setting it elsewhere.

4 years agohttpd/async: remove EINTR check
Eric Wong [Mon, 24 Jun 2019 02:52:23 +0000 (02:52 +0000)]
httpd/async: remove EINTR check

This pipe is always non-blocking when run under public-inbox-httpd
and it won't fail with EINTR in that case

4 years agods: get rid of event_watch field
Eric Wong [Mon, 24 Jun 2019 02:52:22 +0000 (02:52 +0000)]
ds: get rid of event_watch field

We don't need to keep track of that field since we always
know what events we're interested in when using one-shot
wakeups.

4 years agods: remove IO::Poll support (for now)
Eric Wong [Mon, 24 Jun 2019 02:52:21 +0000 (02:52 +0000)]
ds: remove IO::Poll support (for now)

It may be reinstated at a later time if there's interest; but I
want to be able to use one-shot notifications for certain events
while retaining level-triggered notifications others.

OTOH, I intend to fully support kqueue; via IO::KQueue for now,
but via syscall() eventually to take advantage of the syscall
reduction kevent(2) can provide over (current) epoll APIs.

4 years agods: share watch_chg between watch_read/watch_write
Eric Wong [Mon, 24 Jun 2019 02:52:20 +0000 (02:52 +0000)]
ds: share watch_chg between watch_read/watch_write

There was much duplicate logic between watch_read and
watch_write.  Share that logic, and give us room to enable
edge-triggered or one-shot notifications in the future.

4 years agods: import IO::KQueue namespace
Eric Wong [Mon, 24 Jun 2019 02:52:19 +0000 (02:52 +0000)]
ds: import IO::KQueue namespace

Make the rest of our IO::KQueue-using code less verbose and
closer to the C equivalent.

4 years agods: set event flags directly at initialization
Eric Wong [Mon, 24 Jun 2019 02:52:18 +0000 (02:52 +0000)]
ds: set event flags directly at initialization

We can avoid the EPOLL_CTL_ADD && EPOLL_CTL_MOD sequence with
a single EPOLL_CTL_ADD.

4 years agosyscall: get rid of unnecessary uname local vars
Eric Wong [Mon, 24 Jun 2019 02:52:17 +0000 (02:52 +0000)]
syscall: get rid of unnecessary uname local vars

We don't need to keep information from uname(2) around outside
of startup.

4 years agosyscall: get rid of unused EPOLL* constants
Eric Wong [Mon, 24 Jun 2019 02:52:16 +0000 (02:52 +0000)]
syscall: get rid of unused EPOLL* constants

EPOLLRDBAND is used for DECnet; and I'm pretty sure I won't be
updating any of our code to work with DECnet.

I've never found use for EPOLLHUP or EPOLLERR, either; so
disable those for now and add comments for things I might
actually use: EPOLLET and EPOLLONESHOT.

4 years agods: get rid of redundant and unnecessary POLL* constants
Eric Wong [Mon, 24 Jun 2019 02:52:15 +0000 (02:52 +0000)]
ds: get rid of redundant and unnecessary POLL* constants

EPOLL* constants already match their POLL* counterparts and
there's no way Linux can ever diverge or change the values
of those constants.  So we'll favor the EPOLL* ones since we
use EPOLLEXCLUSIVE, already.

For weird stuff like kqueue, we'd need to keep maintaining
the mapping, anyways.

4 years agods: switch write buffering to use a tempfile
Eric Wong [Mon, 24 Jun 2019 02:52:14 +0000 (02:52 +0000)]
ds: switch write buffering to use a tempfile

Data which can't fit into a generously-sized socket buffer,
has no business being stored in heap.

4 years agods: share send(..., MSG_MORE) logic
Eric Wong [Mon, 24 Jun 2019 02:52:13 +0000 (02:52 +0000)]
ds: share send(..., MSG_MORE) logic

No sense in having similar Linux-specific functionality in
both our NNTP.pm and HTTP.pm

4 years agohttp: favor DS->write(strref) when reasonable
Eric Wong [Mon, 24 Jun 2019 02:52:12 +0000 (02:52 +0000)]
http: favor DS->write(strref) when reasonable

This can avoid large memory copies when strings can't be
copy-on-write and saves us the trouble of creating new
refs in the code.

4 years agods: remove support for DS->write(undef)
Eric Wong [Mon, 24 Jun 2019 02:52:11 +0000 (02:52 +0000)]
ds: remove support for DS->write(undef)

We call ->flush_write directly, now; so we can eliminate a
needless check.

4 years agods: don't pass `events' arg to EPOLL_CTL_DEL
Eric Wong [Mon, 24 Jun 2019 02:52:10 +0000 (02:52 +0000)]
ds: don't pass `events' arg to EPOLL_CTL_DEL

There's no point in passing a mask of interesting events
when removing an item from the epoll watch set.

4 years agods: lazy-initialize wbuf
Eric Wong [Mon, 24 Jun 2019 02:52:09 +0000 (02:52 +0000)]
ds: lazy-initialize wbuf

We don't need write buffering unless we encounter slow clients
requesting large responses.  So don't waste a hash slot or
(empty) arrayref for it.

4 years agods: split out from ->flush_write and ->write
Eric Wong [Mon, 24 Jun 2019 02:52:08 +0000 (02:52 +0000)]
ds: split out from ->flush_write and ->write

Get rid of the confusing $need_queue variable and all
the associated documentation for it.  Instead, make it
obvious that we're either skipping the write buffer or
flushing the write buffer by splitting the sub in two.

4 years agods: lazy initialize wbuf_off
Eric Wong [Mon, 24 Jun 2019 02:52:07 +0000 (02:52 +0000)]
ds: lazy initialize wbuf_off

Since Perl 5.10+, "fields" makes a restricted hash; not a
compile-time-defined array (struct) with fixed offsets as
it did in Perl <= 5.8.

Thus in-use fields cost memory, and since the write buffer
offset is rarely needed; stop relying on it.

4 years agods: get rid of on_incomplete_write wrapper
Eric Wong [Mon, 24 Jun 2019 02:52:06 +0000 (02:52 +0000)]
ds: get rid of on_incomplete_write wrapper

Wrong place to be wrapping this method.

4 years agoAddTimer: avoid clock_gettime for the '0' case
Eric Wong [Mon, 24 Jun 2019 02:52:05 +0000 (02:52 +0000)]
AddTimer: avoid clock_gettime for the '0' case

We rely on immediate timers often, so we can avoid the overhead
of an extra subroutine call to retrieve the monotonic time (and
a sometimes-system call on some platforms).

4 years agods: use and export monotonic now()
Eric Wong [Mon, 24 Jun 2019 02:52:04 +0000 (02:52 +0000)]
ds: use and export monotonic now()

All of our internal timing code should use monotonic clocks
for consistency against system clock adjustments.

This can be shared by our Daemon and NNTP packages.

4 years agods: get rid of more unused debug instance methods
Eric Wong [Mon, 24 Jun 2019 02:52:03 +0000 (02:52 +0000)]
ds: get rid of more unused debug instance methods

Over a decade of using Danga::Socket and I never found the
built-in debug functionality useful.

4 years agods: get rid of {closed} field
Eric Wong [Mon, 24 Jun 2019 02:52:02 +0000 (02:52 +0000)]
ds: get rid of {closed} field

Merely checking the presence of the {sock} field is
enough, and having multiple sources of truth increases
confusion and the likelyhood of bugs.

4 years agot/httpd-corner: ensure chunk payload read doesn't overreach
Eric Wong [Thu, 20 Jun 2019 01:03:47 +0000 (01:03 +0000)]
t/httpd-corner: ensure chunk payload read doesn't overreach

It never has, AFAIK, but I'm making some changes to this code in
another branch and nearly introduced a bug where it would be
overreading and discarding the pipelined request.

4 years agot/httpd-corner.t: fix braino :x
Eric Wong [Thu, 20 Jun 2019 00:16:17 +0000 (00:16 +0000)]
t/httpd-corner.t: fix braino :x

Plack is for Perl, Rack is for Ruby; this a Perl project :x

4 years agods: stop distinguishing event read and write callbacks
Eric Wong [Sun, 16 Jun 2019 06:11:28 +0000 (06:11 +0000)]
ds: stop distinguishing event read and write callbacks

Having separate read/write callbacks in every class is too
confusing to my easily-confused mind.  Instead, give every class
an "event_step" callback which is easier to wrap my head around.

This will make future code to support IO::Socket::SSL-wrapped
sockets easier-to-digest, since SSL_write() can require waiting
on POLLIN events, and SSL_read() can require waiting on POLLOUT
events.

4 years agot/replace.t: fix SKIP label for testing w/o Xapian
Eric Wong [Sun, 16 Jun 2019 03:22:25 +0000 (03:22 +0000)]
t/replace.t: fix SKIP label for testing w/o Xapian

4 years agoxcpdb: don't warn on --jobs != --reshard
Eric Wong [Sun, 16 Jun 2019 01:32:01 +0000 (01:32 +0000)]
xcpdb: don't warn on --jobs != --reshard

It's slightly confusing since we dedicate one job
to dealing with fast-import + SQLite indexing; and
it's not worth complaining about when it happens.

4 years agoMerge remote-tracking branch 'origin/newspeak' into xcpdb
Eric Wong [Sun, 16 Jun 2019 01:04:28 +0000 (01:04 +0000)]
Merge remote-tracking branch 'origin/newspeak' into xcpdb

* origin/newspeak:
  comments: replace "partition" with "shard"
  t/xcpdb-reshard: use 'shard' term in local variables
  xapcmd: favor 'shard' over 'part' in local variables
  search: use "shard" for local variable
  v2writable: use "epoch" consistently when referring to git repos
  adminedit: "part" => "shard" for local variables
  v2writable: rename local vars to match Xapian terminology
  v2writable: avoid "part" in internal subs and fields
  search*: rename {partition} => {shard}
  xapcmd: update comments referencing "partitions"
  v2: rename SearchIdxPart => SearchIdxShard
  inboxwritable: s/partitions/shards/ in local var
  tests: change messages to use "shard" instead of partition
  v2writable: rename {partitions} field to {shards}
  v2writable: count_partitions => count_shards
  searchidxpart: start using "shard" in user-visible places
  rename reference to git epochs as "partitions"
  admin|xapcmd: user-facing messages say "shard"
  v2writable: update comments regarding xcpdb --reshard
  doc: rename our Xapian "partitions" to "shards"

4 years agot/psgi_search.t: use higher-level APIs
Eric Wong [Sun, 16 Jun 2019 00:57:36 +0000 (00:57 +0000)]
t/psgi_search.t: use higher-level APIs

No point in using lower-level APIs for a PSGI test.

4 years agosearchview: add link at bottom to reverse results
Eric Wong [Sat, 15 Jun 2019 20:23:42 +0000 (20:23 +0000)]
searchview: add link at bottom to reverse results

I could not find a place to put the link the top without
making navigation too cluttered.  Putting it at the bottom
of the page seems reasonable...

4 years agosearchview: support negative offsets to reverse ordering
Eric Wong [Sat, 15 Jun 2019 20:23:41 +0000 (20:23 +0000)]
searchview: support negative offsets to reverse ordering

Taking a hint from Perl array access, we'll allow negative
offsets for the 'o' parameter and to reverse the sort order.

4 years agoMerge remote-tracking branch 'origin/ds'
Eric Wong [Sat, 15 Jun 2019 17:38:42 +0000 (17:38 +0000)]
Merge remote-tracking branch 'origin/ds'

* origin/ds:
  ds: stop caring about event flags set by epoll/poll/kqueue
  ds: do not distinguish between POLLHUP and POLLERR
  ds: remove read method, here, too
  nntp: use sysread to append to existing buffer
  ds: remove steal_socket method
  ds: remove {fd} field
  ds: reduce Errno imports and drop ->close reason
  ds: cleanup Errno imports and favor constant comparisons
  ds: simplify write buffer accounting

4 years agot/git-http-backend: explain purpose of test
Eric Wong [Fri, 14 Jun 2019 10:09:13 +0000 (10:09 +0000)]
t/git-http-backend: explain purpose of test

I found myself tempted to switch to HTTP::Tiny, here, since
it's distributed with Perl since 5.14, unlike Net::HTTP
(which AFAIK was never a part of Perl proper).

But we really want to use Net::HTTP, here, since it's
lower-level and allows us to trigger server-side buffering
by not reading the entity body.

4 years agocomments: replace "partition" with "shard"
Eric Wong [Sat, 15 Jun 2019 08:25:44 +0000 (08:25 +0000)]
comments: replace "partition" with "shard"

Now that the code matches Xapian terminology, ensure
our comments match, too.

4 years agot/xcpdb-reshard: use 'shard' term in local variables
Eric Wong [Fri, 14 Jun 2019 21:30:31 +0000 (21:30 +0000)]
t/xcpdb-reshard: use 'shard' term in local variables

Another step in maintaining consistency with Xapian docs.

4 years agoxapcmd: favor 'shard' over 'part' in local variables
Eric Wong [Fri, 14 Jun 2019 21:29:37 +0000 (21:29 +0000)]
xapcmd: favor 'shard' over 'part' in local variables

Yet another step to keeping our naming consistent with Xapian
terminology.

4 years agosearch: use "shard" for local variable
Eric Wong [Fri, 14 Jun 2019 18:18:06 +0000 (18:18 +0000)]
search: use "shard" for local variable

Another small step towards terminology consistency with Xapian.

4 years agov2writable: use "epoch" consistently when referring to git repos
Eric Wong [Fri, 14 Jun 2019 18:16:09 +0000 (18:16 +0000)]
v2writable: use "epoch" consistently when referring to git repos

Be consistent with our own terminology and use "epoch" for
[0-9]+\.git repos.  The term "partition" is going away entirely.

4 years agoadminedit: "part" => "shard" for local variables
Eric Wong [Fri, 14 Jun 2019 18:14:42 +0000 (18:14 +0000)]
adminedit: "part" => "shard" for local variables

4 years agov2writable: rename local vars to match Xapian terminology
Eric Wong [Fri, 14 Jun 2019 18:00:20 +0000 (18:00 +0000)]
v2writable: rename local vars to match Xapian terminology

4 years agov2writable: avoid "part" in internal subs and fields
Eric Wong [Fri, 14 Jun 2019 17:50:45 +0000 (17:50 +0000)]
v2writable: avoid "part" in internal subs and fields

We'll be using the term "shard" from now on to be consistent
with Xapian terminology.

4 years agosearch*: rename {partition} => {shard}
Eric Wong [Fri, 14 Jun 2019 17:42:57 +0000 (17:42 +0000)]
search*: rename {partition} => {shard}

Another step towards keeping our internal data structures
consistent with Xapian naming.

4 years agoxapcmd: update comments referencing "partitions"
Eric Wong [Fri, 14 Jun 2019 17:38:01 +0000 (17:38 +0000)]
xapcmd: update comments referencing "partitions"

Don't confuse future readers of our code.

4 years agov2: rename SearchIdxPart => SearchIdxShard
Eric Wong [Fri, 14 Jun 2019 17:35:04 +0000 (17:35 +0000)]
v2: rename SearchIdxPart => SearchIdxShard

Another step towards keeping our file and package names
consistent with Xapian terminology.

4 years agoinboxwritable: s/partitions/shards/ in local var
Eric Wong [Fri, 14 Jun 2019 17:31:28 +0000 (17:31 +0000)]
inboxwritable: s/partitions/shards/ in local var

More work towards being consistent with Xapian's own terminology

4 years agotests: change messages to use "shard" instead of partition
Eric Wong [Fri, 14 Jun 2019 08:02:32 +0000 (08:02 +0000)]
tests: change messages to use "shard" instead of partition

Another potentially user-facing piece made consistent with
Xapian terminology.

4 years agov2writable: rename {partitions} field to {shards}
Eric Wong [Fri, 14 Jun 2019 07:59:53 +0000 (07:59 +0000)]
v2writable: rename {partitions} field to {shards}

Our internal data structure should be consistent with Xapian
terminology.

4 years agov2writable: count_partitions => count_shards
Eric Wong [Fri, 14 Jun 2019 07:56:58 +0000 (07:56 +0000)]
v2writable: count_partitions => count_shards

Another step towards becoming consistent with Xapian terminology

4 years agosearchidxpart: start using "shard" in user-visible places
Eric Wong [Fri, 14 Jun 2019 07:55:35 +0000 (07:55 +0000)]
searchidxpart: start using "shard" in user-visible places

We'll name our process title with "shard" instead, and
update a few error messages and comments to match.

4 years agorename reference to git epochs as "partitions"
Eric Wong [Fri, 14 Jun 2019 07:51:45 +0000 (07:51 +0000)]
rename reference to git epochs as "partitions"

Try to remain consistent with our own documentation regarding
v2 git "epochs", first.

4 years agoadmin|xapcmd: user-facing messages say "shard"
Eric Wong [Fri, 14 Jun 2019 07:37:03 +0000 (07:37 +0000)]
admin|xapcmd: user-facing messages say "shard"

We're slowly getting rid of the word "partition" when it
comes to remain consistent with Xapian docs.

4 years agov2writable: update comments regarding xcpdb --reshard
Eric Wong [Fri, 14 Jun 2019 07:34:07 +0000 (07:34 +0000)]
v2writable: update comments regarding xcpdb --reshard

Using compact to change shard count was abandoned during
the v2 development phase.

4 years agodoc: rename our Xapian "partitions" to "shards"
Eric Wong [Fri, 14 Jun 2019 06:38:58 +0000 (06:38 +0000)]
doc: rename our Xapian "partitions" to "shards"

For consistency with Xapian documentation (in the "master"
branch).

4 years agov2writable: fix brainfart when capping NPROC_MAX_DEFAULT
Eric Wong [Fri, 14 Jun 2019 21:43:16 +0000 (21:43 +0000)]
v2writable: fix brainfart when capping NPROC_MAX_DEFAULT

Oops :x

4 years agosearch: require PublicInbox::Inbox ref here
Eric Wong [Fri, 14 Jun 2019 17:13:04 +0000 (17:13 +0000)]
search: require PublicInbox::Inbox ref here

No sense in supporting multiple methods of initialization
for an internal class.

4 years agosearchidx: require PublicInbox::Inbox (or InboxWritable) ref
Eric Wong [Fri, 14 Jun 2019 17:01:55 +0000 (17:01 +0000)]
searchidx: require PublicInbox::Inbox (or InboxWritable) ref

PublicInbox::Inbox objects have minimal dependencies, so
drop code to support old tests which existed before the
PublicInbox::Inbox object came into existence.

4 years agoMakefile.PL: add test scripts to syntax checks
Eric Wong [Fri, 14 Jun 2019 16:47:44 +0000 (16:47 +0000)]
Makefile.PL: add test scripts to syntax checks

I make syntax errors all the time :x

4 years agot/www_listing: favor HTTP::Tiny over Net::HTTP
Eric Wong [Fri, 14 Jun 2019 16:42:26 +0000 (16:42 +0000)]
t/www_listing: favor HTTP::Tiny over Net::HTTP

More testers are likely to have HTTP::Tiny than Net::HTTP, since
HTTP::Tiny is a dual-life module and distributed with Perl since
Perl 5.14 (2011-05-14), whereas Net::HTTP will likely live in
a separate package forever.

4 years agoedit: fix portability of editor invocation
Eric Wong [Mon, 10 Jun 2019 23:53:46 +0000 (23:53 +0000)]
edit: fix portability of editor invocation

The eval was unnecessary, and $0 can't be "--".
Tested with /bin/sh on FreeBSD 11.2

4 years agoMerge remote-tracking branch 'origin/reshard' into next
Eric Wong [Fri, 14 Jun 2019 16:25:39 +0000 (16:25 +0000)]
Merge remote-tracking branch 'origin/reshard' into next

* origin/reshard:
  xcpdb: support resharding v2 repos
  xcpdb: use destination shard as progress prefix
  xapcmd: preserve indexlevel based on the destination
  v2writable: use a smaller default for Xapian partitions

4 years agoMerge remote-tracking branch 'origin/manifest' into next
Eric Wong [Fri, 14 Jun 2019 16:23:13 +0000 (16:23 +0000)]
Merge remote-tracking branch 'origin/manifest' into next

* origin/manifest:
  git: ensure ->modified returns an integer
  www: support $INBOX/git/$EPOCH.git for v2 cloning
  www: wire up /$INBOX/manifest.js.gz, too
  wwwlisting: generate grokmirror-compatible manifest.js.gz
  wwwlisting: allow hiding entries from manifest

4 years agoMerge remote-tracking branch 'origin/edit' into next
Eric Wong [Fri, 14 Jun 2019 16:23:00 +0000 (16:23 +0000)]
Merge remote-tracking branch 'origin/edit' into next

* origin/edit:
  edit: unlink temporary file when done
  v2writable: replace: kill git processes before reindexing
  edit: drop unwanted headers before noop check
  edit|purge: improve output on rewrites
  edit: new tool to perform edits
  doc: document the --prune option for -index
  admin: expose ->config
  AdminEdit: move editability checks from -purge
  admin: beef up resolve_inboxes to handle purge options
  purge: start moving common options to AdminEdit module
  admin: remove warning arg for unconfigured inboxes
  v2writable: implement ->replace call
  import: switch to "replace_oids" interface for purge
  import: extract_author_info becomes extract_commit_info
  v2writable: consolidate overview and indexing call

4 years agoxcpdb: support resharding v2 repos
Eric Wong [Thu, 13 Jun 2019 00:29:37 +0000 (00:29 +0000)]
xcpdb: support resharding v2 repos

v2 repos are sometimes created on machines where CPU
parallelization exceeds the capability of the storage devices.

In that case, users may reshard the Xapian DB to any smaller,
positive integer to avoid excessive overhead and contention when
bottlenecked by slow storage.

Resharding can also be used to increase shard count after
hardware upgrades.

4 years agoxcpdb: use destination shard as progress prefix
Eric Wong [Wed, 12 Jun 2019 08:10:04 +0000 (08:10 +0000)]
xcpdb: use destination shard as progress prefix

For M:N resharding, we'll want to display the number from
the new shard number.

4 years agoxapcmd: preserve indexlevel based on the destination
Eric Wong [Wed, 12 Jun 2019 08:05:33 +0000 (08:05 +0000)]
xapcmd: preserve indexlevel based on the destination

To support M:N resharding, we need to ensure we store the
indexlevel in the destination shard, rather than the
originating one.

4 years agov2writable: use a smaller default for Xapian partitions
Eric Wong [Wed, 12 Jun 2019 00:35:32 +0000 (00:35 +0000)]
v2writable: use a smaller default for Xapian partitions

Apparently 16 CPUs (probably HT) and SATA storage is common
these days.  Having excessive Xapian partitions leads to
contention and excessive FD/space use.  So set a smaller
default but continue allowing user-specified values to bump
this up.

4 years agogit: remove cat_file sub callback interface
Eric Wong [Thu, 13 Jun 2019 08:10:02 +0000 (08:10 +0000)]
git: remove cat_file sub callback interface

We weren't using it, and in retrospect, it makes no sense to use
this API cat_file for giant responses which can't read quickly
with minimal context-switching (or sanely fit into memory for
Email::Simple/Email::MIME).

For giant blobs which we don't want slurped in memory, we'll
spawn a short-lived git-cat-file process like we do in ViewVCS.

Otherwise, monopolizing a git-cat-file process for a giant
blob is harmful to other PSGI/NNTP users.

A better interface is coming which will be more suitable for
for batch processing of "small" objects such as commits and
email blobs.

4 years agonntp: filter out duplicate Message-IDs for leafnode
Eric Wong [Fri, 14 Jun 2019 00:27:31 +0000 (00:27 +0000)]
nntp: filter out duplicate Message-IDs for leafnode

It's the unfortunate reality that there are some clients which
reuse Message-IDs (in which we generate + use another) or set
multiple Message-IDs on their own.  While the v2 format
addresses that, NNTP clients such as leafnode are not always
prepared to deal with that case.

So, ensure NNTP clients only see a single Message-ID, and
show the others as 'X-Alt-Message-ID'.

4 years agonntp: ensure Message-ID is not folded for leafnode
Eric Wong [Thu, 13 Jun 2019 20:46:52 +0000 (20:46 +0000)]
nntp: ensure Message-ID is not folded for leafnode

Leafnode cannot handle Message-ID headers which are too long and
require folding via Email::Simple::Header.  Since there are
already many of these messages in git with the header already
folded, we need to handle the unfolding when emitting the
message via NNTP.

As far as we know, Leafnode is the only client software
incapable of handling this case.

4 years agonntp: add Path: header for leafnode
Eric Wong [Thu, 13 Jun 2019 06:54:13 +0000 (06:54 +0000)]
nntp: add Path: header for leafnode

Apparently leafnode just needs any junk in the Path: header.
Lets not waste bandwidth and just use a single byte to keep
leafnode happy.

Cc: Dave Taht <dave@taht.net>
4 years agoMakefile.PL: "dsyn" target to check syntax of changed files
Eric Wong [Thu, 13 Jun 2019 03:44:41 +0000 (03:44 +0000)]
Makefile.PL: "dsyn" target to check syntax of changed files

We have lots of files and syntax-checking every single one of
them is slow.  Enable "perl -w" in the existing syntax check
while we're at it.

4 years agodoc: update dependencies for CentOS 7.x
Eric Wong [Thu, 13 Jun 2019 02:26:30 +0000 (02:26 +0000)]
doc: update dependencies for CentOS 7.x

Digest::SHA is the most notable missing package at runtime
for a minimal system.

Tests don't run at all without Test::Simple (or Test::More).
Plack::Test is also a separate package, too...

Also, the package for IO::Compress::Gzip should be IO::Compress;
as perl-PerlIO-gzip is a different thing entirely which is not
relevant to our needs.

Test::HTTP::Server::Simple doesn't seem required at all for Plack
tests.

ExtUtils::MakeMaker needs to be documented as a install dependency
for people installing this, too; since AFAIK public-inbox is not
yet in any distros.