git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH] cvsimport: keep one index per branch during import

From
MLMartin Langhoff <martin@catalyst.net.nz>
Date
Jun 12, 2006, 11:50 UTC
Message-ID
<11501130491187-git-send-email-martin@catalyst.net.nz>

With this patch we have a speedup and much lower IO when importing trees with many branches. Instead of forcing index re-population for each branch switch, we keep many index files around, one per branch.

Signed-off-by: Martin Langhoff <martin@catalyst.net.nz>
---

This patch should get some review. It is trivial, but not fully tested. I am testing it on the moz repo (which will take a while) to check that I get the same result with and without it.

Performance-wise, it seems to be doing ~15K commits per hour, with the mozilla repo, up from ~6Kcph on the same hardware. Of course, this is only noticeable in projects with lots of concurrent branches. Linear projects don't get much from this patch.

With this change, we are now truly waiting on cvs to hand over the files pronto! Running locally, it is apparent that it isn't IO wait but the latency of the chatty cvs protocol that is making this slow.

Probably forking 2 or 3 processes to prefetch filerevs from cvs
and put them in a queue directory for the main process to pick
up would work wonders. Actually, they could call git-hash-object
and just put some file metadata in the queue directory. 
---
 git-cvsimport.perl |   37 ++++++++++++++++++++++++++++++-------
 1 files changed, 30 insertions(+), 7 deletions(-)
diff --git a/git-cvsimport.perl b/git-cvsimport.perl
old mode 100755
new mode 100644
index 76f6246..9c4588f
--- a/git-cvsimport.perl
+++ b/git-cvsimport.perl
@@ -465,10 +465,15 @@ my $git_dir = $ENV{"GIT_DIR"} || ".git";
 $ENV{"GIT_DIR"} = $git_dir;
 my $orig_git_index;
 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
-my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
-				    DIR => File::Spec->tmpdir());
-close ($git_ih);
-$ENV{GIT_INDEX_FILE} = $git_index;
+
+my %index; # holds filenames of one index per branch
+{   # init with an index for origin
+    my ($fh, $fn) = tempfile('gitXXXXXX', SUFFIX => '.idx',
+			     DIR => File::Spec->tmpdir());
+    close ($fh);
+    $index{$opt_o} = $fn;
+}
+$ENV{GIT_INDEX_FILE} = $index{$opt_o};
 unless(-d $git_dir) {
 	system("git-init-db");
 	die "Cannot init the GIT db at $git_tree: $?\n" if $?;
@@ -496,6 +501,13 @@ unless(-d $git_dir) {
 	$tip_at_start = `git-rev-parse --verify HEAD`;
 
 	# populate index
+	unless ($index{$last_branch}) {
+	    my ($fh, $fn) = tempfile('gitXXXXXX', SUFFIX => '.idx',
+				     DIR => File::Spec->tmpdir());
+	    close ($fh);
+	    $index{$last_branch} = $fn;
+	}
+	$ENV{GIT_INDEX_FILE} = $index{$last_branch};
 	system('git-read-tree', $last_branch);
 	die "read-tree failed: $?\n" if $?;
 
@@ -776,8 +788,17 @@ while(<CVS>) {
 		}
 		if(($ancestor || $branch) ne $last_branch) {
 			print "Switching from $last_branch to $branch\n" if $opt_v;
-			system("git-read-tree", $branch);
-			die "read-tree failed: $?\n" if $?;
+			unless ($index{$branch}) {
+			    my ($fh, $fn) = tempfile('gitXXXXXX', SUFFIX => '.idx',
+						     DIR => File::Spec->tmpdir());
+			    close ($fh);
+			    $index{$branch} = $fn;
+			    $ENV{GIT_INDEX_FILE} = $index{$branch};
+			    system("git-read-tree", $branch);
+			    die "read-tree failed: $?\n" if $?;
+			} else {
+			    $ENV{GIT_INDEX_FILE} = $index{$branch};
+		        }
 		}
 		$last_branch = $branch if $branch ne $last_branch;
 		$state = 9;
@@ -841,7 +862,9 @@ #	VERSION:1.96->1.96.2.1
 }
 commit() if $branch and $state != 11;
 
-unlink($git_index);
+foreach my $git_index (values %index) {
+    unlink($git_index);
+}
 
 if (defined $orig_git_index) {
 	$ENV{GIT_INDEX_FILE} = $orig_git_index;
-- 
1.4.0.g5fba
Message 1 of 1 in “cvsimport: keep one index per branch during import”
  1. cvsimport: keep one index per branch during importMartin Langhoff, Jun 12, 2006

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.