Re: MTD git repository status

[Thomas Gleixner <tglx@linutronix.de>]

On Monday 07 November 2005 19:15, Thomas Gleixner wrote:
> I've updated the mtd git repository hopefully in time for 2.6.15.

Some rant about MTD CVS commits

I asked some time ago for some discipline regarding the CVS usage.

Some MTD developers gracefully provide

- separate commits for non related changes
- patches without trailing white spaces
- syntactically correct commit messages
- understandable short and long comments

	Thanks


Some others use the CVS down to their whim.

Is it so hard to comply to some simple rules ?

1. Commit after testing and do not abuse the CVS as your private trash bin

If you have the urge to document your hacking sessions, do this in your 
private CVS or create a branch in the MTD-CVS preferably named 
mtd-yourname-tinkerbox

2. Check for trailing white space _before_ you commit.

3. Do _not_ commit changes which add 
   a)	// old code or
	/* old code */
   b)	#if 0
		old code
	#endif

   a) is a no no
   b) use when such an exclude is necessary:
	/* Comment why this is disabled */
	#ifdef ENABLE_BROKEN_FEATURE
	#endif

4. Commit unrelated changes in separate commits

5. Use the requested commit message format
   [MTD] <core, chips, maps, nand, onenand> <file>: Short description
   or
   [JFFS2] Short description
   <Blank line>
   <Long description>	

6. Write understandable comments, which give an information what the commit is 
doing, rather than some sloppy blurb. Using cryptic abbreviations is also not 
really helpful. See also linux/Documentation/SubmittingPatches


Some example commits for illustration:

---------------------------------------------------------

[JFFS2] minor refinements
[JFFS2] remove forgotten stuff

Really descriptive !

---------------------------------------------------------

Add support for "4G Systems MTX-1 Flash device", better known as meshcube.
Modified Files:
        Kconfig Makefile.common mphysmap.c
...

This is obviously a patch for non kernel parts of the CVS. Err, wait....

Depends mphysmap.c on MTX-1 support or is it the other way round ?

---------------------------------------------------------

[JFFS2] Fix debugging stuff here as well

Here ? In Timbuktu ?

---------------------------------------------------------

Commits with commit messages consisting of 3 empty lines

Is this a guessing game ?

---------------------------------------------------------

[JFFS2] roll-back the last fix. JFFS2 uses strange technique - upper callers 
clean the shit which the lower function leave... In theory, everybody should 
free its own crap... Would be nice to fix - i believe, not everything is 
cleaned well.

Excellent short description! Perfectly fitting in a line of 80 chars. And the 
careful wording is extraordinary.

It definitely leads to extended eruptions of swearing when somebody has to 
review and clean up that #§$%&@*!

---------------------------------------------------------

Five consecutive commits changing one file forth and back within 10 minutes !

Really simple to review, especially when the commit messages are so clear and 
the commit flood changes random files aside the one which is the main working 
target.

---------------------------------------------------------

[OneNAND] More fancy handling of DDP
<END of commit message>

Is [OneNAND] a new subsystem ?

DDP ? 
I know what DDP means in this context, but when somebody else tries to 
understand the message, he might try Google first:

Some Google explanantions from the first page
Delivery Duty Paid
Distributed Data Processing
Data Delivery Protocol
Decentralized Development Planning

---------------------------------------------------------

Trailing white spaces introduced since the last git update

diffstat whitespace.diff
 drivers/mtd/Kconfig                 |    4 -
 drivers/mtd/chips/cfi_cmdset_0001.c |    6 -
 drivers/mtd/inftlmount.c            |    2
 drivers/mtd/maps/ixp2000.c          |    2
 drivers/mtd/maps/mtx-1_flash.c      |    2
 drivers/mtd/maps/pq2fads.c          |   14 +--
 drivers/mtd/maps/tqm834x.c          |    8 +-
 drivers/mtd/maps/ts5500_flash.c     |    2
 drivers/mtd/nand/au1550nd.c         |    6 -
 drivers/mtd/nand/s3c2410.c          |    2
 drivers/mtd/nand/sharpsl.c          |    6 -
 drivers/mtd/onenand/onenand_base.c  |   24 +++---
 drivers/mtd/rfd_ftl.c               |  112 ++++++++++++++---------------
 fs/jffs2/build.c                    |    2
 fs/jffs2/debug.c                    |   28 +++----
 fs/jffs2/debug.h                    |    8 +-
 fs/jffs2/gc.c                       |    4 -
 fs/jffs2/nodelist.c                 |  136 
++++++++++++++++++------------------
 fs/jffs2/nodemgmt.c                 |    6 -
 fs/jffs2/readinode.c                |   64 ++++++++--------
 fs/jffs2/summary.c                  |    8 +-
 fs/jffs2/wbuf.c                     |   14 +--
 include/linux/jffs2.h               |    2
 include/linux/jffs2_fs_sb.h         |    2
 include/linux/mtd/bbm.h             |    2
 include/linux/mtd/cfi.h             |    2
 include/linux/mtd/onenand.h         |    2
 27 files changed, 235 insertions(+), 235 deletions(-)

Figure yourself, who earns the most credits for being sloppy

	tglx