public-inbox.git
2 years agopublic-inbox 1.6.0 v1.6.0
Eric Wong [Wed, 16 Sep 2020 20:02:20 +0000 (20:02 +0000)]
public-inbox 1.6.0

2 years agogit_async_cat: fix outdated comment
Eric Wong [Wed, 16 Sep 2020 07:18:53 +0000 (07:18 +0000)]
git_async_cat: fix outdated comment

We replaced Danga::Socket with PublicInbox::DS roughly a year
before GitAsyncCat was introduced into our git history.

2 years agowwwtext: link to public-inbox.org/meta archives
Eric Wong [Tue, 15 Sep 2020 19:48:50 +0000 (19:48 +0000)]
wwwtext: link to public-inbox.org/meta archives

Since we're advertising our address at meta@public-inbox.org,
we should advertise the archives, too.

2 years agowwwstream: link to cgit URLs for coderepo
Eric Wong [Tue, 15 Sep 2020 20:15:24 +0000 (20:15 +0000)]
wwwstream: link to cgit URLs for coderepo

Hopefully this reduces the ambiguity between code for the
project(s) using public-inbox and the code for public-inbox
itself.

2 years agotreewide: relax allow >=40 chars for git OID
Eric Wong [Tue, 15 Sep 2020 19:51:37 +0000 (19:51 +0000)]
treewide: relax allow >=40 chars for git OID

This will help with eventual git SHA-256 transitions.

2 years agomid: rename MID_MAX to ID_MAX
Eric Wong [Tue, 15 Sep 2020 19:51:36 +0000 (19:51 +0000)]
mid: rename MID_MAX to ID_MAX

It's only used for HTML anchors which we will need indefinitely.

2 years agoimap: quiet uninitialized variable warning on FETCH
Eric Wong [Tue, 15 Sep 2020 06:11:18 +0000 (06:11 +0000)]
imap: quiet uninitialized variable warning on FETCH

This was triggered by blindly trying to FETCH an MSN (not
"UID FETCH") on an empty dummy inbox.  It's harmless, and
probably triggered by a wayward client or misbehaving bot.

2 years agoci/deps: add Plack::Test::ExternalServer for devtest
Eric Wong [Mon, 14 Sep 2020 22:02:23 +0000 (22:02 +0000)]
ci/deps: add Plack::Test::ExternalServer for devtest

More of our Plack tests exercise public-inbox-httpd, nowadays;
and ExternalServer lets us test it easily alongside generic PSGI
stuff.

2 years agot/imapd.t: skip dependent test on failure
Eric Wong [Mon, 14 Sep 2020 21:29:40 +0000 (21:29 +0000)]
t/imapd.t: skip dependent test on failure

We don't want to cascade failures/warnings when something else
breaks.  There's likely more of these to be fixed as we
encounter them.

2 years agodoc: TODO and release notes updates ahead of 1.6
Eric Wong [Mon, 14 Sep 2020 06:39:36 +0000 (06:39 +0000)]
doc: TODO and release notes updates ahead of 1.6

Some more things have happened...

And drop some items which are too expensive to support,
such as automatic mirroring.

2 years agotests: consistently check for xapian-compact
Eric Wong [Mon, 14 Sep 2020 06:29:26 +0000 (06:29 +0000)]
tests: consistently check for xapian-compact

We may need to test against development versions of Xapian,
which may rely on setting `XAPIAN_COMPACT=xapian-compact-1.5'.
Ensure it's possible to do that.

And add a missing check in t/xcpdb-reshard.t, too.

2 years agosigfd: fix typos and scoping on systems w/o epoll+kqueue
Eric Wong [Mon, 14 Sep 2020 03:42:30 +0000 (03:42 +0000)]
sigfd: fix typos and scoping on systems w/o epoll+kqueue

Unfortunately, I'm not sure how easy catching these at
compile-time, is.  Prototypes do not seem to check these
at compile time when crossing packages (not even with
exported subroutines).

2 years agodoc: Add piem to list of clients
Kyle Meyer [Sun, 13 Sep 2020 19:31:55 +0000 (15:31 -0400)]
doc: Add piem to list of clients

2 years agonntp: share more code between art_lookup callers
Eric Wong [Fri, 11 Sep 2020 07:32:33 +0000 (07:32 +0000)]
nntp: share more code between art_lookup callers

This prepares us for future changes to improve scalability to
many inboxes.

2 years agot/nntpd: add test for the XPATH command
Eric Wong [Fri, 11 Sep 2020 07:32:32 +0000 (07:32 +0000)]
t/nntpd: add test for the XPATH command

It's only in RFC 2980 (not 977 or 3977), but Net::NNTP has
supported it since 2001, at least.  We'll be making changes
to avoid pathological behavior, so test it, first.

2 years agotreewide: avoid `goto &NAME' for tail recursion
Eric Wong [Fri, 11 Sep 2020 07:32:31 +0000 (07:32 +0000)]
treewide: avoid `goto &NAME' for tail recursion

While Perl implements tail recursion via `goto' which allows
avoiding warnings on deep recursion.  It doesn't (as of 5.28)
optimize the speed of such dispatches, though it may reduce
ephemeral memory usage.

Make the code less alien to hackers coming from other languages
by using normal subroutine dispatch.  It's actually slightly
faster in micro benchmarks due to the complexity of `goto &NAME'.

2 years agowwwstream: show init + index instructions for -V1, too
Eric Wong [Thu, 10 Sep 2020 05:46:04 +0000 (05:46 +0000)]
wwwstream: show init + index instructions for -V1, too

This should've always been there.  I'm not sure how widely
spread 1.0 and earlier releases were, but we'll keep documenting
the version requirement.

2 years agosolver: async blob retrieval for diff extraction
Eric Wong [Thu, 10 Sep 2020 01:51:53 +0000 (01:51 +0000)]
solver: async blob retrieval for diff extraction

Like the rest of the WWW code, public-inbox-httpd now uses
git_async_cat to retrieve blobs without blocking the event loop.
This improves fairness when git blobs are on slow storage and
allows us to take better advantage of SMP systems.

2 years agosolver: break apart inbox blob retrieval
Eric Wong [Wed, 9 Sep 2020 06:26:18 +0000 (06:26 +0000)]
solver: break apart inbox blob retrieval

To avoid hogging the event loop in public-inbox-httpd when
many candidate messages match, we'll separate the steps to
ensure fairness on slow storage.

2 years agosolver: check one git coderepo and inbox at a time
Eric Wong [Wed, 9 Sep 2020 06:26:17 +0000 (06:26 +0000)]
solver: check one git coderepo and inbox at a time

With public-inbox-httpd, this mitigates the effect of slow git
blob storage with multiple coderepos configured for an inbox.
It's still synchronous for now (and may need to remain that way
for ->last_check_err), but no longer monopolizes the event loop
when checking multiple coderepos.

We don't yet support multi-inbox scanning, yet; but this also
prepares us for a future where we do.

We'll also support >=40 char blob OIDs in preparation for future
git SHA-256 support, too.

2 years agowwwlisting: avoid hogging event loop
Eric Wong [Wed, 9 Sep 2020 06:26:16 +0000 (06:26 +0000)]
wwwlisting: avoid hogging event loop

By using the just-introduced ConfigIter class.
And make ManifestJsGz a subclass of it to reduce duplication.

2 years agoextmsg: prevent cross-inbox matches from hogging event loop
Eric Wong [Wed, 9 Sep 2020 06:26:15 +0000 (06:26 +0000)]
extmsg: prevent cross-inbox matches from hogging event loop

With many inboxes, checking multiple SQLite repos will be slow
and time-consuming, so ensure we can schedule it fairly between
multiple inboxes.

2 years agot/cgi.t: show stderr on failures
Eric Wong [Wed, 9 Sep 2020 06:26:14 +0000 (06:26 +0000)]
t/cgi.t: show stderr on failures

This helped me diagnose an error I would've introduced
in the next commit.

2 years agoconfig: split out iterator into separate object
Eric Wong [Wed, 9 Sep 2020 06:26:13 +0000 (06:26 +0000)]
config: split out iterator into separate object

We will need to allow simultaneous iterators on the same
config object, since we'll need this for ExtMsg, NNTPD,
WwwListing, NewsWWW, and other places.

2 years agoconfig: flatten each_inbox and iterate_start args
Eric Wong [Wed, 9 Sep 2020 06:26:12 +0000 (06:26 +0000)]
config: flatten each_inbox and iterate_start args

In Perl, we can simplify callers by passing a single array
all the way down the stack instead of a single array ref which
needs to be expanded every call.

2 years agowww: manifest.js.gz generation no longer hogs event loop
Eric Wong [Wed, 9 Sep 2020 06:26:11 +0000 (06:26 +0000)]
www: manifest.js.gz generation no longer hogs event loop

It's still as slow as before with hundreds/thousands of inboxes,
but at least it's fair.  Future changes will allow it to be
cached and memoized with persistent HTTP servers.

2 years agouse "\&" where possible when referring to subroutines
Eric Wong [Wed, 9 Sep 2020 06:26:10 +0000 (06:26 +0000)]
use "\&" where possible when referring to subroutines

"*foo" is ambiguous in that it may refer to a bareword file handle;
so we'll use it where we can without triggering warnings.

PublicInbox::TestCommon::run_script_exit required dropping the
prototype, however.  We'll also future-proof by dropping "use
warnings" in Cgit.pm and use the less-ambiguous "//=" in Inbox.pm
while we're in the area.

2 years agosolver: drop warnings, modernize use v5.10.1, use SEEK_SET
Eric Wong [Wed, 9 Sep 2020 06:26:09 +0000 (06:26 +0000)]
solver: drop warnings, modernize use v5.10.1, use SEEK_SET

With Perl upstream preparing to deprecate things, we'll move
towards only enabling warnings during development via shebang
and stop enabling them via "use".

We'll also favor "use v5.10.1" over the Perl 5.6-compatible "use
5.010_001", since our code base never worked on 5.6.

Finally, were also importing SEEK_SET without using it, just use it
for readability since we can't avoid loading Fcntl in other
places and it'll get constant-folded, anyways.

2 years agoxt/solver: test with public-inbox-httpd, too
Eric Wong [Wed, 9 Sep 2020 06:26:08 +0000 (06:26 +0000)]
xt/solver: test with public-inbox-httpd, too

We'll be making changes to solver to make it even fairer
to slow clients on slow storage.  Ensure we test with
public-inbox-httpd-specific codepaths, since the generic
PSGI code paths are rare in production use.

2 years agowwwtext: config comment improvements
Eric Wong [Wed, 9 Sep 2020 21:23:26 +0000 (21:23 +0000)]
wwwtext: config comment improvements

Use the full URL of the inbox being mirrored to reduce ambiguity
(instead of just the inbox name).

Using asymmetric quotes (e.g `foo') improves readability for me
in that it's more obvious when a quote begins and ends.  It also
lights up fewer pixels and reduces visual noise compared to
double-quotes.

We'll also reflow the `mainrepo' vs `inboxdir' comment slightly
to emphasize the word `instead'.

2 years agowwwtext: don't blindly quote "git clone" destination
Eric Wong [Wed, 9 Sep 2020 21:23:25 +0000 (21:23 +0000)]
wwwtext: don't blindly quote "git clone" destination

Save screen space and light up fewer pixels to reduce visual noise.

2 years agowwwtext: describe the use of `coderepo' entries
Eric Wong [Wed, 9 Sep 2020 21:23:24 +0000 (21:23 +0000)]
wwwtext: describe the use of `coderepo' entries

The `solver' feature is not very obvious, give potential
users a hint about it.

2 years agonntp: fix cross-newsgroup Message-ID lookups
Eric Wong [Thu, 10 Sep 2020 09:38:39 +0000 (09:38 +0000)]
nntp: fix cross-newsgroup Message-ID lookups

We cannot blindly use the selected newsgroup for
HEAD/ARTICLE/BODY requests using Message-ID, since
those commands look across all newsgroups; not just
the selected one (if any).

So stuff a reference to the Inbox object into $smsg.
We can reduce args passed into set_nntp_headers() and
msg_hdr_write(), too.

Fixes: 0e6ceff37fc38f28 ("nntp: support slow blob retrievals")
2 years agowwwstream: fix "Atom feed" link
Eric Wong [Wed, 9 Sep 2020 21:38:17 +0000 (21:38 +0000)]
wwwstream: fix "Atom feed" link

Oops, I wanted to stop escaping double-quotes with `qq()' but
used `q()' instead :x

Fixes: 2f61828fcb727e51 ("www: make mirror instructions more prominent")
2 years agocontrib/css: limit <a> coloring to links, only
Eric Wong [Wed, 9 Sep 2020 20:24:55 +0000 (20:24 +0000)]
contrib/css: limit <a> coloring to links, only

We don't want <a> tags without href= attributes to be colored,
since the `<a id=mirror>' tag in the HTML footer is intended
as an anchor destination for `<a href=#mirror>' link at the
top.

2 years agowww: make mirror instructions more prominent
Eric Wong [Tue, 8 Sep 2020 08:29:14 +0000 (08:29 +0000)]
www: make mirror instructions more prominent

In order to fight the misconception that public-inboxes are
centralized, anchor "#mirror" to the clone instructions and
place an emphasis on "mirror", not just cloning.

While we're at it, better describe multi-epoch -V2 inboxes,
since some users do not seem to realize epochs consist of
different data.

2 years agov2writable: reuse read-only shard counting code
Eric Wong [Wed, 2 Sep 2020 11:04:21 +0000 (11:04 +0000)]
v2writable: reuse read-only shard counting code

We'll also fix the read-only code to ensure we notice missing
Xapian shards, since gaps would throw off our expectation that
Xapian document IDs and NNTP article numbers are interchangeable.

2 years agooveridx: document column uses
Eric Wong [Wed, 2 Sep 2020 11:04:20 +0000 (11:04 +0000)]
overidx: document column uses

This may be useful for keeping our heads on straight dealing
with IMAP, NNTP, JMAP, etc.

2 years agowwwaltid: drop unused sqlite3_missing function
Eric Wong [Wed, 2 Sep 2020 11:04:19 +0000 (11:04 +0000)]
wwwaltid: drop unused sqlite3_missing function

It's inlined into the main function, which we'll shorten
slightly with the defined-or (`//') operator.  Also noticed
and fixed a mismatched HTML tag.

2 years agoimap: drop old, pre-Parse::RecDescent search parser
Eric Wong [Wed, 2 Sep 2020 11:04:18 +0000 (11:04 +0000)]
imap: drop old, pre-Parse::RecDescent search parser

We switched to Parse::RecDescent during development and left
some dead code behind.

2 years agosearch: remove {over_ro} field
Eric Wong [Wed, 2 Sep 2020 11:04:17 +0000 (11:04 +0000)]
search: remove {over_ro} field

Only inbox accesses the read-only {over}, now, instead of going
through ->search.  This simplifies our object graph and avoids
potentially redundant FDs and DB handles pointing to the same
over.sqlite3 file.

2 years agosearch: replace ->query with ->mset
Eric Wong [Wed, 2 Sep 2020 11:04:16 +0000 (11:04 +0000)]
search: replace ->query with ->mset

Nearly all of the search uses in the production code rely on
a Xapian mset iterator being returned (instead of an array
of $smsg objects).  So default to returning the mset and move
the burden of smsg array conversion into the test cases.

2 years agotests: add "use strict" and declare v5.10.1 compatibility
Eric Wong [Wed, 2 Sep 2020 11:04:15 +0000 (11:04 +0000)]
tests: add "use strict" and declare v5.10.1 compatibility

strict.pm helped me find a typo in an upcoming recent change, so
ensure we use it since it does more good than harm.  We'll also
take the opportunity here to declare v5.10.1 compatibility level
to future-proof against Perl incompatibilities.

2 years agosearch: remove special case for blank query
Eric Wong [Wed, 2 Sep 2020 11:04:14 +0000 (11:04 +0000)]
search: remove special case for blank query

The special case (if any) belongs at a higher-level,
and this is another step towards removing {over_ro}-dependence
in our Search object.

2 years agouse more idiomatic internal API for ->over access
Eric Wong [Wed, 2 Sep 2020 11:04:13 +0000 (11:04 +0000)]
use more idiomatic internal API for ->over access

{over_ro} being a part of the Search object is a historical
oddity which will go away, soon.  Lets start removing its use in
tests and rarely-used helper scripts.

2 years agodisambiguate OverIdx and Over by field name
Eric Wong [Wed, 2 Sep 2020 11:04:12 +0000 (11:04 +0000)]
disambiguate OverIdx and Over by field name

We'll use {oidx} as the common field name for the read-write
OverIdx, here, to disambiguate it from the read-only {over}
field.  This hopefully makes it clearer which code paths are
read-only and which are read-write.

2 years agomsgmap: note how we use ->created_at
Eric Wong [Wed, 2 Sep 2020 11:04:11 +0000 (11:04 +0000)]
msgmap: note how we use ->created_at

It'll likely be used in the future for JMAP, detached indices,
and maybe other things.

2 years agot/run: Perl future proofing
Eric Wong [Mon, 31 Aug 2020 23:41:56 +0000 (23:41 +0000)]
t/run: Perl future proofing

Bareword file handles outside of STD(IN|OUT|ERR) seem to be on
the chopping block for Perl 8.  We'll also "use v5.10.1" to
guard against future incompatibilities.

2 years agoinit+convert: create non-existing directory hierarchies
Eric Wong [Tue, 1 Sep 2020 01:15:07 +0000 (01:15 +0000)]
init+convert: create non-existing directory hierarchies

Following "git init" as an example, we'll create every parent
path up to the one specified, instead of attempting to continue
on when Cwd::abs_path returns `undef'.

2 years agodoc: remove B<> (bold) markup from the remaining POD
Eric Wong [Tue, 1 Sep 2020 01:15:06 +0000 (01:15 +0000)]
doc: remove B<> (bold) markup from the remaining POD

B<> decreases readability of the POD source and is of dubious
usefulness in the man page.

2 years agowatch: add --help/-h support
Eric Wong [Tue, 1 Sep 2020 01:15:05 +0000 (01:15 +0000)]
watch: add --help/-h support

And avoid unnecessary POD markup in the man page.

2 years agoconfig: use defined-or (//) in a few places
Eric Wong [Tue, 1 Sep 2020 01:15:04 +0000 (01:15 +0000)]
config: use defined-or (//) in a few places

Just some golfing to reduce scrolling and hopefully readability.

2 years agomda+learn: add --help / -h support
Eric Wong [Tue, 1 Sep 2020 01:15:03 +0000 (01:15 +0000)]
mda+learn: add --help / -h support

"use Getopt::Long" doesn't seem too slow on a hot page cache,
and it's probably used frequently enough to be in cache.

We'll also start reducing the amount of markup in the .pod and
favoring verbatim text in documentation for readability in
source form, since the bold text seems excessive.

2 years agodaemon: support --help/-h in -httpd/imapd/nntpd
Eric Wong [Tue, 1 Sep 2020 01:15:02 +0000 (01:15 +0000)]
daemon: support --help/-h in -httpd/imapd/nntpd

For consistency with other commands, though the
protocol-specific options should refer users to
the manpage.

2 years agoscript/*: fold $usage into $help, support `-h' instead of -?
Eric Wong [Tue, 1 Sep 2020 01:15:01 +0000 (01:15 +0000)]
script/*: fold $usage into $help, support `-h' instead of -?

`-h' doesn't conflict with anything, and some users (including
git users) may be more accustomed to using it rather than the
rarely-seen-outside-of-Getopt::Long `-?' switch.

We can also rely on the GetOptions() function to emit a proper
error message instead of just "bad command-line args".

2 years agoedit+purge: support `--help' and `-h' like other commands
Eric Wong [Tue, 1 Sep 2020 01:15:00 +0000 (01:15 +0000)]
edit+purge: support `--help' and `-h' like other commands

And while we're at it, note edit is *destructive* to encourage
reading the fine manual.

2 years agoadmin: improve minimum version text
Eric Wong [Tue, 1 Sep 2020 01:14:59 +0000 (01:14 +0000)]
admin: improve minimum version text

"inboxes 1 inboxes not supported by ..." was non-sensical.
Now it'll show "-V1 inbox not supported by ...", instead.

2 years agoscript/*: set executable bit on -learn and -imapd
Eric Wong [Tue, 1 Sep 2020 01:14:58 +0000 (01:14 +0000)]
script/*: set executable bit on -learn and -imapd

It's useful to mark they're meant to be executable, even
if the shebang is useless.

2 years agot/v2dupindex: test indexing mirrors with duplicate messages
Eric Wong [Tue, 1 Sep 2020 05:55:45 +0000 (05:55 +0000)]
t/v2dupindex: test indexing mirrors with duplicate messages

While it's not a known problem, our deduplicating logic may
change in the future; or a BOFH could be manually injecting
duplicate messages directly into the git epoch repositories.

Ensure indexing in mirrors doesn't break when there's
duplicates.  This is in preparation for detached indices
for multi-inbox search.

2 years agoindex: check for xapian-compact when using --compact
Eric Wong [Tue, 1 Sep 2020 16:54:31 +0000 (16:54 +0000)]
index: check for xapian-compact when using --compact

Otherwise, users may be frustrated to discover it missing
a long indexing run.

2 years agoreplace ParentPipe with EOFpipe
Eric Wong [Mon, 31 Aug 2020 04:41:40 +0000 (04:41 +0000)]
replace ParentPipe with EOFpipe

ParentPipe was a subset of EOFpipe, except EOFpipe correctly
accounts for theoretical(*) spurious wakeups on the pipe.

(*) AFAIK, spurious wakeups are/were more likely on TCP sockets
    due to checksum failures, something that's not a problem on
    local pipes.  We're also not sharing pipes like we do with
    listen sockets on accept(2), so there's no chance of another
    process grabbing bytes (unless we have bugs in our code).

2 years agods: avoid unnecessary timer for waitpid
Eric Wong [Mon, 31 Aug 2020 04:41:39 +0000 (04:41 +0000)]
ds: avoid unnecessary timer for waitpid

It doesn't seem necessary, since we won't call dwaitpid()
until we see an EOF.

2 years agowatch: use EOFpipe to reduce dwaitpid wakeups
Eric Wong [Mon, 31 Aug 2020 04:41:38 +0000 (04:41 +0000)]
watch: use EOFpipe to reduce dwaitpid wakeups

It's a bit inefficient to use a pipe, here.  However, using
dwaitpid() on a process that's not expected to exit soon is
also inefficient as it causes excessive wakeups as most of
our inbox-writing code expects synchronous waitpid().

This only affects -watch instances configured for NNTP and IMAP
clients.

2 years agods: avoid excessive queueing when reaping PIDs
Eric Wong [Mon, 31 Aug 2020 04:41:37 +0000 (04:41 +0000)]
ds: avoid excessive queueing when reaping PIDs

We should not enqueue reap_pids() to run more than once per
EventLoop iteration.  We'll start reformatting reap_pids
to tabs, too, since we're no longer Danga::Socket.

We should also be able to remove timer usage for reaping
down-the-line once we stop abusing dwaitpid() in -watch.

2 years agowatch: comments and tiny cleanups
Eric Wong [Mon, 31 Aug 2020 04:41:36 +0000 (04:41 +0000)]
watch: comments and tiny cleanups

Get rid of an unused variable, prefix a warning and try to
better document control flow around various callbacks.

2 years agowatch: block signals before fork on non-signalfd/kevent systems
Eric Wong [Mon, 31 Aug 2020 04:41:35 +0000 (04:41 +0000)]
watch: block signals before fork on non-signalfd/kevent systems

In case there's non-Linux or BSD users w/o IO::KQueue, we
shouldn't let signal handlers fire in the child processes.

The child processes always assumed signals were blocked by
the parent, so no changes were necessary, there.

2 years agowatch: avoid unnecessary spawning on spam removals
Eric Wong [Mon, 31 Aug 2020 04:41:34 +0000 (04:41 +0000)]
watch: avoid unnecessary spawning on spam removals

This should further mitigate lock contention problems
when -watch is configured to watch on a Maildir for spam
while performing a large NNTP import.

There is now a small risk a message won't get removed because if
it's in the current (uncommitted) fast-import batch, but
unlikely given the batch size is now only 10 messages.

If a that small window is hit, flipping the \Seen flag
(e.g. marking it unread, and then read again) will trigger
another removal attempt via IMAP or Maildir.

2 years agowatch: log signal activities to STDERR
Eric Wong [Mon, 31 Aug 2020 04:41:33 +0000 (04:41 +0000)]
watch: log signal activities to STDERR

Sometimes it may not be apparent when/if a signal is
processed, this hopefully improves the situation.

We'll also change the process title when we're quitting
to better inform users.

2 years agorename WatchMaildir => Watch
Eric Wong [Mon, 31 Aug 2020 04:41:32 +0000 (04:41 +0000)]
rename WatchMaildir => Watch

This is no longer limited to Maildirs now that IMAP and NNTP
support exist; so give it a shorter name.

2 years agowatchmaildir: use v5.10.1, drop warnings
Eric Wong [Mon, 31 Aug 2020 04:41:31 +0000 (04:41 +0000)]
watchmaildir: use v5.10.1, drop warnings

Declare 5.10.1 to avoid potential compatibility problems with
Perl 7/8 down the line.  We'll rely on the command-line to set
or drop warnings during development, at least.

2 years agowatch: limit batch size of NNTP and IMAP workers, too
Eric Wong [Mon, 31 Aug 2020 04:41:30 +0000 (04:41 +0000)]
watch: limit batch size of NNTP and IMAP workers, too

We don't want to monopolize locks because processes can easily
block each other if using `watchspam' on a Maildir while a big
NNTP or IMAP import is happening.

This can also happen if somebody configured a single inbox to
watch from several sources to merge several mailboxes into one
(e.g. both an IMAP and Maildir are watched).

2 years agodoc: expand on indexBatchSize regarding fragmentation
Eric Wong [Mon, 31 Aug 2020 04:33:37 +0000 (04:33 +0000)]
doc: expand on indexBatchSize regarding fragmentation

And change the documentation reference in -tuning to
point to the -index manpage while we're at it.

2 years agoimapd: filter out unusable flags from search
Eric Wong [Sat, 29 Aug 2020 20:32:19 +0000 (20:32 +0000)]
imapd: filter out unusable flags from search

Quiet down logs from -imapd when clients are blindly
sending some unsupported flag conditions (e.g. "DRAFT",
"DELETED") specified in RFC 3501.

2 years agotests: check-run: fixup un-squashed simplification
Eric Wong [Sat, 29 Aug 2020 03:48:39 +0000 (03:48 +0000)]
tests: check-run: fixup un-squashed simplification

Link: https://public-inbox.org/meta/20200828221803.GA89978@dcvr/
2 years agotests: check-run: show skipped tests
Eric Wong [Fri, 28 Aug 2020 10:13:00 +0000 (10:13 +0000)]
tests: check-run: show skipped tests

We'll deduplicate redundant lines and show counts of skipped
tests to ensure it's easy to notice if something is unexpectedly
skipped.

2 years agoimaptracker: update_last: simplify callers
Eric Wong [Fri, 28 Aug 2020 10:12:59 +0000 (10:12 +0000)]
imaptracker: update_last: simplify callers

By making it a no-op if last_uid is not defined.  This isn't a
hot code path, so the extra method dispatch isn't an issue.
It'll save some indentation/wrapping in future commits.

2 years agowatch: flush changes to inbox before updating IMAPTracker
Eric Wong [Fri, 28 Aug 2020 10:12:58 +0000 (10:12 +0000)]
watch: flush changes to inbox before updating IMAPTracker

Data needs to hit inboxes, first.  Otherwise it's possible to
skip messages in case git-fast-import is killed before it sees
"done\n".  Now, -watch will just waste a little bandwidth in
re-downloading a seen message if it's interrupted immediately
before updating IMAPTracker.

2 years agoMakefile.PL: run check-man for <= 80 columns on check-run, too
Eric Wong [Fri, 28 Aug 2020 04:22:00 +0000 (04:22 +0000)]
Makefile.PL: run check-man for <= 80 columns on check-run, too

I mostly use "make check-run" instead of the slower "make check"
target, nowadays, so add this check to ensure the rendered
manpage is always be visible to more users who need big fonts.

2 years agowww: more descriptive pagination
Eric Wong [Thu, 27 Aug 2020 22:05:00 +0000 (22:05 +0000)]
www: more descriptive pagination

Being an easily confused person, I find "next" and "prev"
ambiguous as to whether messages on the next or previous page
will be newer or older than the current page.  Clarify that for
the threaded /$INBOX/ view and search results.

For search results sorted by relevance, we'll use "[>= $SCORE]"
or "[<= $SCORE]" to indicate to indicate directionality.

This also fixes $INBOX/new.html for unindexed v1 inboxes.

2 years agowww: improve navigation around contemporary threads
Eric Wong [Thu, 27 Aug 2020 22:04:59 +0000 (22:04 +0000)]
www: improve navigation around contemporary threads

Sometimes it's useful to quickly get to threads and messages
which are contemporaries of the current thread/message being
focused on.  This hopefully improves navigation by making:

a) the top line (where $INBOX_DIR/description) is shown
   a link to the latest topics in search results and
   per-thread/per-message views.

b) providing a link to contemporaries ("~YYYY-MM-DD") at
   around the thread overview skeleton area for per-thread
   and per-message views

2 years agodoc: watch: expand on NNTP and IMAP-specific knobs
Eric Wong [Thu, 27 Aug 2020 12:17:06 +0000 (12:17 +0000)]
doc: watch: expand on NNTP and IMAP-specific knobs

There's a few more, but maybe they're too esoteric
to be worth documenting at the moment (batch sizes, timeouts, etc).

2 years agodoc: move watch config docs to -watch manpage
Eric Wong [Thu, 27 Aug 2020 12:17:05 +0000 (12:17 +0000)]
doc: move watch config docs to -watch manpage

The -config manpage is a bit long and the -watch stuff is
isolated from the rest of it while we start documenting NNTP and
IMAP support.

I'm not entirely happy with the way IMAP and NNTP are
configured, it's still good enough for small setups.

This also fixes a long-standing misplaced comment about
`publicinboxwatch.spamcheck' affecting all configured inboxes,
that comment was actually for `publicinboxwatch.watchspam'.

We'll omit documenting NNTP for `watchspam', for now, given the
lack of \Seen flags in NNTP and I'm not sure if it's even
useful.  There may not be any newsgroups for sharing confirmed
spam, either...

2 years agowatch: imap: only remove \Seen spam
Eric Wong [Thu, 27 Aug 2020 12:17:04 +0000 (12:17 +0000)]
watch: imap: only remove \Seen spam

This matches the behavior of Maildir `watchspam' handling in not
removing unseen messages.  NNTP can't match this behavior, since
NNTP servers don't store flags, clients do.

2 years agodoc: speling fickses
Eric Wong [Thu, 27 Aug 2020 12:17:03 +0000 (12:17 +0000)]
doc: speling fickses

2 years agodoc: document graceful shutdown signals
Eric Wong [Thu, 27 Aug 2020 12:17:02 +0000 (12:17 +0000)]
doc: document graceful shutdown signals

Same as the read-only daemons.

2 years agooveridx: inline create_ghost sub
Eric Wong [Thu, 27 Aug 2020 12:17:01 +0000 (12:17 +0000)]
overidx: inline create_ghost sub

There's no need for this to be a separate sub since there's
only a single caller.  This saves a few kilobytes at least
in short-lived processes.

2 years agoimaptracker: preserve WAL journal_mode if set by user
Eric Wong [Thu, 27 Aug 2020 12:17:00 +0000 (12:17 +0000)]
imaptracker: preserve WAL journal_mode if set by user

It's no problem for most users to enable WAL, here, since
there's only a single process doing both reading and writing
(unlike the read-only daemons).  However, WAL doesn't work on
network filesystems, so it can't be enabled by default.

2 years agowatchmaildir: ensure I:/W:/E: prefixes in warnings
Eric Wong [Thu, 27 Aug 2020 12:16:59 +0000 (12:16 +0000)]
watchmaildir: ensure I:/W:/E: prefixes in warnings

For consistency in output, any URL/path-context-dependent
prefixes should have the same prefix as the actual warning which
triggered it.

2 years agogit: show more context info on failures
Eric Wong [Thu, 27 Aug 2020 07:51:25 +0000 (07:51 +0000)]
git: show more context info on failures

I'm seeing "read: Connection timed out" from in my syslog from
-httpd.  The fail() calls in PublicInbox::Git seems to be the
only code path of ours which could trigger it...

ETIMEDOUT shouldn't happen on pipes, only sockets; and all of
our socket operations are non-blocking.  So this could be
cgit-wwwhighlight-filter.lua, but that's connecting over
localhost, though on fairly loaded HW.

2 years agosearch: allow testing with current xapian.git and 1.5.x
Eric Wong [Wed, 26 Aug 2020 22:02:57 +0000 (22:02 +0000)]
search: allow testing with current xapian.git and 1.5.x

A `PI_XAPIAN' environment variable is now exposed for testing
purposes.  We'll also deal with the removal of
`NumberValueRangeProcessor' and use `NumberRangeProcessor'
in its place, but continue favoring the old Search::Xapian
since that's all that's packaged for Debian 10.x stable.

2 years agomsgmap: use v5.10.1
Eric Wong [Wed, 26 Aug 2020 08:17:42 +0000 (08:17 +0000)]
msgmap: use v5.10.1

We use the defined-or (`//', `//=') operators in 5.10,
so require 5.10.1 like the rest of our codebase.  Update
an outdated comment while we're at it.

2 years agoover*: use v5.10.1, drop warnings
Eric Wong [Wed, 26 Aug 2020 08:17:41 +0000 (08:17 +0000)]
over*: use v5.10.1, drop warnings

v5.10.1 lets us use the lighter parent.pm instead of base.pm,
and we'll rely on the shebang to enable warnings (or not).

While we're in the area, drop a no-longer-necessary import for
PublicInbox::Search, since OverIdx doesn't require search.

2 years agoover: recent: remove expensive COUNT query
Eric Wong [Wed, 26 Aug 2020 08:17:40 +0000 (08:17 +0000)]
over: recent: remove expensive COUNT query

As noted in commit 87dca6d8d5988c5eb54019cca342450b0b7dd6b7
("www: rework query responses to avoid COUNT in SQLite"),
COUNT on many rows is expensive on big SQLite DBs.

We've already stopped using that code path long ago in WWW
while -imapd and -nntpd never used it.  So we'll adjust our
remaining test cases to not need it, either.

2 years agoover: rename ->disconnect to ->dbh_close
Eric Wong [Wed, 26 Aug 2020 08:17:39 +0000 (08:17 +0000)]
over: rename ->disconnect to ->dbh_close

Since we got rid of over->connect, `disconnect' no longer pairs
with it.  So name it after the `close(2)' syscall it ultimately
issues.

2 years agoover: rename ->connect method to ->dbh
Eric Wong [Wed, 26 Aug 2020 08:17:38 +0000 (08:17 +0000)]
over: rename ->connect method to ->dbh

`->connect' is confused with the perlfunc for the `connect(2)'
syscall, and also `DBI->connect'.  Since SQLite doesn't use
sockets, the word "connect" needlessly confuses me.  Give
it a short name to match the field name we use for it, which
also matches the variable name used by the DBI(3pm) and
DBD::SQLite(3pm) manpages.

2 years agov2writable: compatibility with SWIG Xapian binding
Eric Wong [Tue, 25 Aug 2020 20:26:24 +0000 (20:26 +0000)]
v2writable: compatibility with SWIG Xapian binding

The SWIG binding won't auto-convert IV/UV to PV like the XS
Search::Xapian binding would, so workaround that shortcoming
for now.

Fixes: a367ec1b15a2458 ("mbox: disable "&t" on existing Xapian until full reindex")
2 years agogrok-pull.post_update_hook: flock(2) before SQLite check
Eric Wong [Tue, 25 Aug 2020 10:23:14 +0000 (10:23 +0000)]
grok-pull.post_update_hook: flock(2) before SQLite check

Unlike DBD::SQLite, the sqlite3(1) CLI does not have a default
busy timeout enabled, so it easily times out while acquiring a
SHARED lock for read-only queries.  We can avoid battery-wasting
polling from the SQLite timeout handler by relying on flock(2)
as we do in our Perl code.

Furthermore, this avoids triggering some locking problems[1]
from a long "SELECT COUNT(*) ..." query and reindex.

While there may be other SQLite-related parallelism issues[1],
this works around one of them by relying on flock(2).

[1] https://public-inbox.org/meta/20200825001204.GA840@dcvr/

2 years agoover+msgmap: respect WAL journal_mode if set
Eric Wong [Tue, 25 Aug 2020 03:02:47 +0000 (03:02 +0000)]
over+msgmap: respect WAL journal_mode if set

WAL actually seems to have ideal locking characteristics given
concurrency problems I'm experiencing with --reindex running
in parallel with expensive read-only SQLite queries:
<https://public-inbox.org/meta/20200825001204.GA840@dcvr/>

Unfortunately, we cannot blindly use WAL while preserving
compatibility with existing setups nor our guarantees that
read-only daemons are indeed "read-only".

However, respect an user's the choice to set WAL on their
own if they're comfortable with giving -nntpd/-httpd/-imapd
processes write permission to the directory storing SQLite DBs.

2 years agomsgmap: use "CREATE TABLE IF NOT EXISTS"
Eric Wong [Tue, 25 Aug 2020 03:02:46 +0000 (03:02 +0000)]
msgmap: use "CREATE TABLE IF NOT EXISTS"

It's fewer queries and matches what we do in OverIdx.

2 years agoover: skip nodatacow on the journal
Eric Wong [Tue, 25 Aug 2020 03:02:45 +0000 (03:02 +0000)]
over: skip nodatacow on the journal

This file gets truncated anyhow, so it won't fragment.