Re: [Squashfs-devel] squashfs xattr support?

[Phillip L. <ph...@lo...>]

Helge Bahmann wrote:
> Am Friday 23 April 2010 11:49:24 schrieben Sie:
>> On Fri, Apr 23, 2010 at 10:04 AM, Helge Bahmann
>>
>>> Or are you too busy (or not interested in the topic) and would prefer
>>> squashfs+xattr to be brought forward without any involvement on your
>>> part?
>> Come on are you being serious here?  Squashfs is a project I created
>> from scratch
>> and have worked on almost non-stop for over 8 years.  So what do you think?
> 
> I think that I have sent you a private mail in Nov 2009 querying about 
> squashfs+xattr, a public mail to this list in March 2010, and another one on 
> Tuesday this week. Neither of these mails has incited a reply on your part, 
> and what I infer from three non-replies is what everybody else is likely to 
> infer as well.
> 

I don't recall the private email in Nov 2009.  Before there's any more horrible
misunderstandings, I'm not denying you sent it, just that I simply can't remember
reading it.

> But since I have apparently been mistaken I offer my apologies, but would like 
> to point out that you could have prevented this misunderstanding by simply 
> replying a bit sooner.
> 
> Let's put it this way: I had been under the impression that you were not going 
> to do xattr any time soon (and I wanted to see it rather sooner than later), 
> so I simply went along to see if I could implement it in 2-3 days. Obviously I 
> could, so for me the problem is essentially solved.
> 

I worked out an xattr layout and I partially implemented it in 2008.  The lack
of a public announcement about the work up till now was because it wasn't ready to
be publicly announced.  Documenting the layout (or releasing the mksquashfs)
more or less commits me to keeping that layout, even if subsequent
performance tests (which obviously can't be done yet, as it's not finished),
show modifications are needed.

I've been caught in this trap before.  I have released (privately to people
who asked for it) pre-releases of new Squashfs file system layouts, only to
find I need to alter the layout following further performance tests
(only it's now too late).  Or I have released a new layout only to find I need
to release a new one a couple of months later (Squashfs 2.0 and Squashfs 2.1
with an enhanced layout is a good example).

People complained repeatedly to me about the succession of new and incompatible
layouts, and so I decided never to release details about a new layout before
I was 100% sure it was correct and was usable for all anticipated purposes for a
couple of years or more.  This has worked well, so over the last 4 years there's
only been two new layouts 3.0 (March 2006) and 4.0 (~September 2008).  In contrast
previously between 2002 and 2006, I might be forced to release a revision to the
layout every year, or more frequently.

So as I said I'm caught in a trap.  Previously I got caught in the mess of
releasing new layouts early to people that wanted it.  Now, I'm reticent about
releasing details, and people think I'm being provocative and malicious.


> Frankly I could therefore just sit here being content that I have a working 
> squashfs+xattr while everybody else hasn't. But since squashfs is your 
> creation and I have drawn great use from this, I am offering you the work I 
> have done in return, you are free to take it or leave it. So far you have not 
> yet offered me the courtesy of a reply on technical matters.
> 

Please see above.  If you want I can comment technically on your layout.  I have
not until now because I didn't want to publicly criticise your layout, without
having posted my layout in return.  This could definately be seen as
provocative and malicious.

>> I care too much about Squashfs to allow a situation where any old crap
>> gets put into Squashfs without my say so.
> 
> I object to my code being classified as "crap".

I apologise.  In no way did I mean to imply your code was crap.

Phillip