threads / patch / 1278

patch, 9 partsapplymbox updates and assorted trivial patches.

Subject: [PATCH 0/9] applymbox updates and assorted trivial patches.

## tl;dr

3 messages between Jul 20, 2005 and Jul 21, 2005. Diffs are folded; open one to read it.

replies: 2people: 2as markdown or json

Junio C Hamano· Jul 20, 2005, 23:30 UTC · lore

I was experimenting with applypatch and applymbox to see how it feels like being Linus ;-) and also I saw some trivially correct patches on the list floating unmerged during KS/OLS.

Since I rebase my repostitory to fix up the history often, I needed a reliable way to preserve authorship information when I export the patches via format-patch and slurp them via applymbox.

I am sending the following changes to applymbox and format-patch first:

  [PATCH] tools/applymbox: allow manual fixing and continuing after a failure.
  [PATCH] git-format-patch-script and mailinfo updates.
  [PATCH] format-patch: --mbox and --check.

I wrote the first one to recover from a situation where an earlier patch failed to apply but I still had many other e-mails in the mailbox.

The second one allows format-patch to give ability to optionally
add From: and Date: lines as the first two lines after the patch
title line.  The mailinfo program has been enhanced to grok
Date: in the top part of the message body just like From: in the
message body can override the sender of the e-mail.

The third one is an extra. It lets me export patches in a format resembling UNIX mbox, so I can concatenate the output together and reprocess them through applymbox.

After they are accepted, I will forward these trivially correct patches I saw on the list, using the enhanced format-patch; they are expected to be processed with the new applymbox as a practice; by doing this I am also ACKing these patches:

  [PATCH] debian/ fixes
  [PATCH] Fix a typo in git-unpack-objects documentation.
  [PATCH] Cleanup: git-verify-tag-script
  [PATCH] Install git-verify-tag-script
  [PATCH] Support more http features: https no cert, .netrc -> auth
  [PATCH] Document "curl" requirements.
Ryan Anderson· Jul 21, 2005, 06:19 UTC · re: Junio C Hamano · lore

Re: [PATCH 0/9] applymbox updates and assorted trivial patches.

On Wed, Jul 20, 2005 at 04:30:29PM -0700, Junio C Hamano wrote:
> Since I rebase my repostitory to fix up the history often, I
> needed a reliable way to preserve authorship information when I
> export the patches via format-patch and slurp them via applymbox.

Is it just me, or does format-patch lose the commit comments during export? I haven't experimented with your new patches yet - and since I'm at OLS and playing with Git during my free time, my brain is pretty much shut down now - so if you have happen to have fixed that, thanks!

[...]
Show 6 quoted lines
> After they are accepted, I will forward these trivially correct
> patches I saw on the list, using the enhanced format-patch; they
> are expected to be processed with the new applymbox as a
> practice; by doing this I am also ACKing these patches:
> 
>   [PATCH] debian/ fixes

I have more debian/ fixes - I cc:ed you on them since I see you are collecting things this week.

-- 
Ryan Anderson
  sometimes Pug Majere
Junio C Hamano· Jul 21, 2005, 06:58 UTC · re: Ryan Anderson · lore

Re: [PATCH 0/9] applymbox updates and assorted trivial patches.

Ryan Anderson <ryan@michonline.com> writes:
> Is it just me, or does format-patch lose the commit comments during
> export?  I haven't experimented with your new patches yet - and since
> I'm at OLS and playing with Git during my free time, my brain is pretty
> much shut down now - so if you have happen to have fixed that, thanks!
Oh, I envy you guys!  Have fun at OLS.

As far as I know, the commit comments are not lost since the original version of format-patch. What I added as extras with today's patch are the original author name, email and author timestamp. There wasn't a provision to record these in the output from format-patch.

The applymbox/applypatch pair Linus did allowed you to have an extra "From:" line as the first line of the e-mail patch submission when you are forwarding a patch somebody else has written. When such e-mail is applied to his repository using applymbox (back then it was called "dotest"), however, the original author timestamp is lost, and instead the commit gets timestamp from the message forwarding the patch. That is what I meant by the loss of information.

Once I acked a patch to Linus with "Author-Date:" line as the second line of the e-mail message (the first being the "From:"), and he told me to refrain from doing it "for now", but that "for now" has never been lifted, so I decided to lift it myself ;-).

The changed format-patch can be told to produce "Date:" (instead of "Author-Date:"), and "From:". There is a corresponding change to the receiving end (the mailinfo program that is used by applymbox/applypatch pair) to read this extra "Date:" line and use that instead of the timestamp obtained from the e-mail submission.

← back to recent threads