[RFC] inbox: add `modified' sub

[RFC] inbox: add `modified' sub

[Eric Wong]

For inboxes with SQLite enabled (all v2, and probably most v1);
stat-ing the msgmap.sqlite3 file is a much faster way of telling
when an inbox is modified compared to scanning git branches.
---
 RFC since I'm not so sure about this one.  It's probably fine,
 but mtime may be worth considering over ctime, too...
 (somebody could be using atomic-rsync on the SQLite and Xapian
  stuff because indexing is slow)

 lib/PublicInbox/Inbox.pm | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/lib/PublicInbox/Inbox.pm b/lib/PublicInbox/Inbox.pm
index cde4625..fa41ab6 100644
--- a/lib/PublicInbox/Inbox.pm
+++ b/lib/PublicInbox/Inbox.pm
@@ -333,4 +333,18 @@ sub recent {
 	search($self)->{over_ro}->recent($opts, $after, $before);
 }
 
+sub modified {
+	my ($self) = @_;
+	my $dir = $self->{mainrepo};
+	my @st;
+	if ($self->{version} >= 2) {
+		@st = stat("$dir/msgmap.sqlite3")
+	} elsif (-f "$dir/public-inbox/msgmap.sqlite3") { # v1
+		@st = stat(_);
+	} else { # v1 repos did not need sqlite3
+		return git($self)->modified;
+	}
+	@st ? $st[10] : time; # 10: ctime
+}
+
 1;
-- 
EW

[PATCH v2] inbox: add `modified' sub

[Eric Wong]

>  but mtime may be worth considering over ctime, too...
>  (somebody could be using atomic-rsync on the SQLite and Xapian
>   stuff because indexing is slow)

-------8<----------
Subject: [PATCH] inbox: add `modified' sub

For inboxes with SQLite enabled (all v2, and probably most v1);
stat-ing the msgmap.sqlite3 file is a much faster way of telling
when an inbox is modified compared to scanning git branches.

v2: use mtime instead of ctime for users who use generic tools
    to copy SQLite DBs.
---
 lib/PublicInbox/Inbox.pm | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/lib/PublicInbox/Inbox.pm b/lib/PublicInbox/Inbox.pm
index cde4625..c004f25 100644
--- a/lib/PublicInbox/Inbox.pm
+++ b/lib/PublicInbox/Inbox.pm
@@ -333,4 +333,20 @@ sub recent {
 	search($self)->{over_ro}->recent($opts, $after, $before);
 }
 
+sub modified {
+	my ($self) = @_;
+	my $dir = $self->{mainrepo};
+	my @st;
+	if ($self->{version} >= 2) {
+		@st = stat("$dir/msgmap.sqlite3")
+	} elsif (-f "$dir/public-inbox/msgmap.sqlite3") { # v1
+		@st = stat(_);
+	} else { # v1 repos did not need sqlite3
+		return git($self)->modified;
+	}
+	# we favor 9/mtime over ctime since it should be possible to
+	# use generic tools such as atomic-rsync to copy SQLite DBs
+	@st ? $st[9] : time;
+}
+
 1;
-- 
EW

[PATCH v3] inbox: add `modified' sub

[Eric Wong]

For inboxes with SQLite enabled (all v2, and probably most v1);
stat-ing the msgmap.sqlite3 file is a much faster way of telling
when an inbox is modified compared to scanning git branches.

v2: use mtime instead of ctime for users who use generic tools
    to copy SQLite DBs.

v3: avoid uninitialized field for v1 repos
    (and I'm not sure why I used the "->{version} || 1" idiom
    everywhere instead of unconditionally setting it in
    Inbox->new...)
---
 lib/PublicInbox/Inbox.pm | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/lib/PublicInbox/Inbox.pm b/lib/PublicInbox/Inbox.pm
index cde4625..ba3d2c7 100644
--- a/lib/PublicInbox/Inbox.pm
+++ b/lib/PublicInbox/Inbox.pm
@@ -333,4 +333,20 @@ sub recent {
 	search($self)->{over_ro}->recent($opts, $after, $before);
 }
 
+sub modified {
+	my ($self) = @_;
+	my $dir = $self->{mainrepo};
+	my @st;
+	if (($self->{version} || 1) >= 2) {
+		@st = stat("$dir/msgmap.sqlite3")
+	} elsif (-f "$dir/public-inbox/msgmap.sqlite3") { # v1
+		@st = stat(_);
+	} else { # v1 repos did not need sqlite3
+		return git($self)->modified;
+	}
+	# we favor 9/mtime over ctime since it should be possible to
+	# use generic tools such as atomic-rsync to copy SQLite DBs
+	@st ? $st[9] : time;
+}
+
 1;
-- 
EW

[PATCH v4] inbox: add `modified' sub

[Eric Wong]

For inboxes with SQLite enabled (all v2, and probably most v1);
we can use the overview DB to get the timestamp of the latest
message.  It's faster than scanning git branches for commit times,
but not always the same.
---
 OK, this is definitely the version I'll go with...
 st_mtime of msgmap.sqlite3 isn't right for mirrors with
 intermittent polling or even if the DB gets rebuilt.

 lib/PublicInbox/Inbox.pm | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/lib/PublicInbox/Inbox.pm b/lib/PublicInbox/Inbox.pm
index cde4625..0d28dd0 100644
--- a/lib/PublicInbox/Inbox.pm
+++ b/lib/PublicInbox/Inbox.pm
@@ -333,4 +333,16 @@ sub recent {
 	search($self)->{over_ro}->recent($opts, $after, $before);
 }
 
+sub modified {
+	my ($self) = @_;
+	if (my $srch = search($self)) {
+		my $msgs = $srch->{over_ro}->recent({limit => 1});
+		if (my $smsg = $msgs->[0]) {
+			return $smsg->{ts};
+		}
+		return time;
+	}
+	git($self)->modified; # v1
+}
+
 1;
-- 
EW