threads / patch / 30036

patchMake 'cvs -n commit ...' not to commit

Subject: [PATCH] Make 'cvs -n commit ...' not to commit

## tl;dr

3 messages between Mar 22, 2012 and Mar 23, 2012. Diffs are folded; open one to read it.

replies: 2people: 2as markdown or json

ericc· Mar 22, 2012, 20:57 UTC · lore

Actually, doing a 'cvs -n commit' will _do_ the commit... With this patch, it now goes through the code, but don't do the commit. A further progress would be to do the pre-commit hook is possible...

Eric Chamberland <Eric.Chamberland@giref.ulaval.ca>
---
 git-cvsserver.perl |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)
Show changes to git-cvsserver.perl +8 −1
diff --git a/git-cvsserver.perl b/git-cvsserver.perl
index b8eddab..67ec4d0 100755
--- a/git-cvsserver.perl
+++ b/git-cvsserver.perl
@@ -1395,6 +1395,9 @@ sub req_ci
         push @committedfiles, $committedfile;
         $log->info("Committing $filename");
 
+        # Don't want to actually _DO_ the update if -n specified
+        unless ( $state->{globaloptions}{-n} ) 
+        {
         system("mkdir","-p",$dirpart) unless ( -d $dirpart );
 
         unless ( $rmflag )
@@ -1424,6 +1427,7 @@ sub req_ci
             $log->info("Updating file '$filename'");
             system("git", "update-index", $filename);
         }
+        }
     }
 
     unless ( scalar(@committedfiles) > 0 )
@@ -1434,6 +1438,9 @@ sub req_ci
         return;
     }
 
+    # Don't want to actually _DO_ the update if -n specified
+    unless ( $state->{globaloptions}{-n} ) 
+    {
     my $treehash = `git write-tree`;
     chomp $treehash;
 
@@ -1537,7 +1544,7 @@ sub req_ci
             print "/$filepart/1.$meta->{revision}//$kopts/\n";
         }
     }
-
+    }
     cleanupWorkTree();
     print "ok\n";
 }

-- 
1.7.3.4
Junio C Hamano· Mar 23, 2012, 18:39 UTC · re: ericc · lore

Re: [PATCH] Make 'cvs -n commit ...' not to commit

ericc <eric.chamberland@giref.ulaval.ca> writes:
> Actually, doing a 'cvs -n commit' will _do_ the commit...
> With this patch, it now goes through the code, but don't do the commit.
OK.
> A further progress would be to do the pre-commit hook is possible...
It is not clear what you meant here.
Show 25 quoted lines
> Eric Chamberland <Eric.Chamberland@giref.ulaval.ca>
> ---
>  git-cvsserver.perl |    9 ++++++++-
>  1 files changed, 8 insertions(+), 1 deletions(-)
>
> diff --git a/git-cvsserver.perl b/git-cvsserver.perl
> index b8eddab..67ec4d0 100755
> --- a/git-cvsserver.perl
> +++ b/git-cvsserver.perl
> @@ -1395,6 +1395,9 @@ sub req_ci
>          push @committedfiles, $committedfile;
>          $log->info("Committing $filename");
>  
> +        # Don't want to actually _DO_ the update if -n specified
> +        unless ( $state->{globaloptions}{-n} ) 
> +        {
>          system("mkdir","-p",$dirpart) unless ( -d $dirpart );
>  
>          unless ( $rmflag )
> @@ -1424,6 +1427,7 @@ sub req_ci
>              $log->info("Updating file '$filename'");
>              system("git", "update-index", $filename);
>          }
> +        }
>      }

I understand that you tried to make the patch smaller by avoiding re-indenting, but this is *yucky*.

It looks to me that the above part could be solved with:
	unless (...) {
		next;
	}

I think the function being patched is too big. Wouldn't it be better to have a refactoring patch to move the above per-path logic to a helper function that deals with a single path, and then insert the "omit call to that helper when run with -n" code in a separate patch?

The same comment applies to the other hunk.

Also I notice that the indentation used throughout the file is somewhat broken (e.g. "Emulate by running hooks/update" part is indented to 8 columns, but earlier parts use 4 space indent). The right structure for this change may be:

 Patch 1: Fix indentation (and do nothing else) to uniformly indent with
          HT;
 Patch 2: Refactor this big funciton using a handful of helper functions
	  (and do nothing else);
 Patch 3: Omit calls to these helper functions under -n option.
Show 19 quoted lines
> @@ -1434,6 +1438,9 @@ sub req_ci
>          return;
>      }
>  
> +    # Don't want to actually _DO_ the update if -n specified
> +    unless ( $state->{globaloptions}{-n} ) 
> +    {
>      my $treehash = `git write-tree`;
>      chomp $treehash;
>  
> @@ -1537,7 +1544,7 @@ sub req_ci
>              print "/$filepart/1.$meta->{revision}//$kopts/\n";
>          }
>      }
> -
> +    }
>      cleanupWorkTree();
>      print "ok\n";
>  }
Eric Chamberland· Mar 23, 2012, 19:02 UTC · re: Junio C Hamano · lore

Re: [PATCH] Make 'cvs -n commit ...' not to commit

On 03/23/2012 02:39 PM, Junio C Hamano wrote:
Show 8 quoted lines
> ericc<eric.chamberland@giref.ulaval.ca>  writes:
>
>> Actually, doing a 'cvs -n commit' will _do_ the commit...
>> With this patch, it now goes through the code, but don't do the commit.
>
> OK.
>
>> A further progress would be to do the pre-commit hook is possible...
Sorry, I wanted to write:
"A further progress would be to do the pre-commit hook *if* possible..."

here, we are used to do "cvs -n commit" just to check if the "hooks" on the cvs server will fail or not...

Show 31 quoted lines
>
> I understand that you tried to make the patch smaller by avoiding
> re-indenting, but this is *yucky*.
>
> It looks to me that the above part could be solved with:
>
> 	unless (...) {
> 		next;
> 	}
>
> I think the function being patched is too big.  Wouldn't it be better to
> have a refactoring patch to move the above per-path logic to a helper
> function that deals with a single path, and then insert the "omit call to
> that helper when run with -n" code in a separate patch?
>
> The same comment applies to the other hunk.
>
> Also I notice that the indentation used throughout the file is somewhat
> broken (e.g. "Emulate by running hooks/update" part is indented to 8
> columns, but earlier parts use 4 space indent).  The right structure for
> this change may be:
>
>   Patch 1: Fix indentation (and do nothing else) to uniformly indent with
>            HT;
>
>   Patch 2: Refactor this big funciton using a handful of helper functions
> 	  (and do nothing else);
>
>   Patch 3: Omit calls to these helper functions under -n option.
>
>

Ok you are right... These were my very first lines in Perl... I just wanted to catch the attention of someone who is able to do the changes correctly... and in a more clean way than I...

Eric

← back to recent threads