[illumos-Discuss] multi-extent files

[Joerg Schilling]

Apostolos Syropoulos <asyropoulos at yahoo.com> wrote:
> in Solaris supports multi-extent  files:
>  
>  http://www.archivum.info/opensolaris-bugs@opensolaris.org/2009-06/00042/%28osol-bugs%29-Question-about-Multi-extent-ISO9660-files.html
>
>  
> When I reported this "bug" Jörg Schilling replied and he wrote in that forum
> that he had prepared everything but SUN did not wanted this additional bits.
> Is it possible that this enhancement will be included in Illumos?

Let me first write some notes on the original plans with Sun....
The original plan was to add Joliet support, to implement support for 
Rock-Ridge-V1.12 and Hardlinks and to add support for large files via 
ISO-9660 multi-extent files. After I implemented Joliet and Hardlink support, 
it took more than 9 months to get a code-review  (although this was a project 
where Sun promised to pay me). This finally resulted in a code integration without 
changes. The delay caused by the Sun burocracy and the unwillingness of the Sun 
employees to do code-reviews for others prevented the project from being able to 
finish before the middle management changed and this new management decided that 
large file support is not needed in Solaris.

Unfortunately it seems that this problem from inside Sun has been imported into 
Illumos... 

>From my current impressions with Illumos, i guess that there is little hope 
that a code integration related to multi-extend support will ever happen 
with Illumos.

What happened before?

In July, Garret D'Amore contacted me for Illumos and made several promises to 
me in order to get me as a supporter for Illumos. None of his promises have 
turned into reality.

In early August (after Illumos has been announced to the public) Garret 
promised to integrate star and soon made some remarks. Some of them have been 
valid and I introduced some changes. Unfortunately, during the past 2 months, I
have been told ad nauseum by him that star integration is a low priority task 
and that Garret is not willing to do a code review that is needed for 
integration. Calling star integration a low priority task sounds a bit strange 
while the official main task for Illumos is to replace closed source and star 
is able to replace closed source in Solaris and star is a mature piece of 
software with no known bugs since two years.

At the same time Garret claims to work on replacing closed source with new OSS 
code. My impression is that this code gets to few testing compared to the 
complexity of the features: I've even seen integrations for software with 
known bugs and without the intention to fix these bugs in the near future.

I explained to Garret that OSS in a community project only works in case that 
people are treated equally and that doing code reviews for others is a mutual 
courtesy. I did even have a 3:15 hour long phone call with him last wednesday 
evening. In this call, I tried to explain which problems I see. At the end of 
that phone call, I had the impression that Garret understood the problems.....
A few hours later, he integrated his 80% ready od(1) replacement code although
it still has several bugs or unacceptable deviations from the POSIX standard.

I tried to warn that Garret's od(1) replacement is in obvious conflict to the 
POSIX standard. Instead of asking for more help, it seems that Garret never 
was interested in help and even claimed that he knows that my statements abut 
his od(1) replacement are wrong.

BTW: I of course offered much more software for integration than just star,
but star is the first integration project and I of course have no interest in
spending my time in other projects that will not be integrated either. Garret 
also made it very clear that he is not interested in software from me as he is
not willing to give my software an equal chance. It seems that I have to give 
up wasting my time in this area (Garret's od(1) replacement currently only
implements 80% of the POSIX requirements correctly and I expect that at least 
60% of the work still has to be put into this implementation to make it 
usable). I like to have a sufficiently comparable OSS replacement in my disto 
"SchilliX" and for this reason Garret's os(1) is not suitable for SchilliX. It
seems to me that I would be more effective if I don't try to help people who 
don't like help me but rather do the work on my own. I am currently writing my
own od(1) replacement and I expect it to be ready before the od(1) 
implementation that is currently in Illumos will offer similar compatibility 
with the Sun implementation and POSIX. If my code is ready, I will of course 
publish it immediately.




Jörg

-- 
 EMail:joerg at schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
       js at cs.tu-berlin.de                (uni)  
       joerg.schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
 URL:  http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily