Autor: Przemyslaw Dobrowolski (dobrawka_at_obta.uw.edu.pl)
Data: Tue 14 Mar 2000 - 17:50:42 MET
Trochę informacji na temat JFS'a. Co do porównywania z HPFS386 miałem te same
odczucia. Ale przeczytajacie, bo jest to bardzo interesujące:
==================BEGIN FORWARDED MESSAGE==================
[...]
>From: Knut.Stange.Osmundsen_at_pmsc.no
>On Tue, 15 Feb 2000 16:01:07 +0100 (CET), Achim Hasenmueller wrote:
>
>>On Tue, 15 Feb 2000 15:05:02 +0100 (CET), Martin Alfredsson wrote:
>>I have been in contact with Dave Kleikamp (lead JFS developer for OS/2
and
>>now Linux) and they do not intend to leave the OS/2 code in the public
>>source tree but they are willing to acknowledge an OS/2 effort and
support
>>it. I was even told that we might get the missing files so that we have
>>a full source tree. If there is a group that really wants to work on JFS,
>>I would be trying to get the missing pieces...
Any success in getting these "missing" pieces? I have had a closer look at
the JFS source. It seems like there are several pieces of new code/bugfixes
within it (later than jfsfix1.zip). (Like F226941 and F226941 in jfs_ifs.c)
Another thing I've noticed is that the JFS.IFS release and jfsfix1.zip
don't have the _JFS_LAZYWRITE defined, so there is no lazywriter thread. I
can't understand why! But this is probably the reason why the write
performance on a JFS volume is quite bad compared with HPFS/HPFS386
volumes. (try a large 'copy' command.) So, it could be quite interesting to
compile a version with a lazywriter thread.
>Why not get the code and open a new section at netlabs.org
>"projects seeking for programmers"
>where you can submit stuff like this and when someone has the
>time and interest he/she could get the code and start coding away.
Good idea! Count me in.
>
>The code could be used to create a FreeJFS but as easily be used
>to create diskutilities for JFS.
>
>In any cases I think we should try to get hold of the sources. The KEE (in
WSEB)
>is said to be undocumented so it might help to peek into the JFS sources
to
>understand the new API's.
With some corrections, the kee.h from Holger Veit seems/is quite usable.
(But I haven't been able to succefully download kee.lib - seems like it's
downloaded as text and not binary, or is simply corrupted.)
>I could hold the code at my site, but I dont know enough (have enough
time) to
>try to build it.
With some creativty creating missing structs and defines, I've succesfully
_compiled_ the existing IFS sources. (I've not been able to link it yet,
due to missing sources and kee.lib)
Regards,
knut
[...]
===================END FORWARDED MESSAGE===================
--- *Przemek/2* - dobrawka_at_asua.org.pl - ICQ: 18642806 Asua Programmers http://www.asua.org.pl TeamOS/2pl http://www.teamos2.org.pl This OS/2 system uptime is 0d 0h 23m 26s 109ms (en).
To archiwum zostało wygenerowane przez hypermail 2.1.7 : Tue 18 May 2004 - 15:26:28 MET DST