>Halfmann, Klaus wrote:
>> I still can not find these offset, can you tell me where in
>> dump1.txt I can find them ?
> 004dbf0: 0000 0000 0000 0000 0ff8 00f8 0078 000e .............x..
>Here's the one for the header node in the catalog tree. The three
>records are at the following offsets:
> r1: 000e (14)
> r2: 0078 (120)
> r3: 00f8 (248)
> Also, the free space begins at 0ff8.
Ahrg, I assumed those offsets at the end of the Catalog _fork_
not at the end of the Catalog _record_ (Ok, I can throw away some
of my code now :). Hmm I miss some start of the Btree map recod
let me study this stupid technote again ... Ok I think I had
a deep mistunderstanding of the hierarchy of structures. As
I understdand the hierarchy now as follows:
Volume (contains some predefined forks)
Nodes (cocantenated via next/prev)
BTrees (spanning over Nodes)
(Btree-)records (files/folders)
Forks of files (data / rsrc) as found in btrees.
Extends file
Btree
extends records (keyed by parent / file cnid)
(The technote really lacks an explanantion like this ...)
The imposes, that one can do a binary search on those keys,
fine.
>> Yes , I just started to grasp whats up with the extends overflow
>> file. I intended to add this functionality later. (Ill come
>> away without for some time I assume :)
>Not a huge deal. Only a heavily fragmented filesystem ever uses
>this. I don't have any HFS+ filesystems where this isn't empty.
Ill try to force MacOS into creating some extends and then
will implement it ...
-> added to my ToDo list in index.html
>> 'LiH0', 'Linu', 'LiH+'
> Sounds fine. I guess I'll have to get some for the kernel as well,
> whenever I actually make it write to the disk.
-> added to my ToDo list in index.html
> Yes, I have a very good net connection at work. I would be able
> to catch the last couple hours of the time you are on that way.
>> Do you receive mails online ? If yes I could send you
>> a mail and upon a direct answer we could go online simultaneously.
> I don't generally look at my personal email while I'm at work, but
> I keep an eye on my work email. (bboyer@yahoo-inc.com)
> I also am almost always logged in with Yahoo! messenger during the
> day as well. I even have a linuxppc build of the client. (It isn't
> on the official download site yet, but it should get added soon.)
> <A HREF="http://messenger.yahoo.com/" TARGET="_blank">
> http://messenger.yahoo.com/</A>
> I apologize for the shameless plug. I work for Yahoo! ;)
Oh well, I earn my money by doing Java programming on NT and
(starting soon) AS400 (Yes you can do Java on those boxes :-)
(Hm, but since they use PowerPC theyd bette'r run ppcLinux :-( )
Well, Ill try this messanger than. In case this might fail
ill 'ping' you at your email adress. I'd suggest we use irc
this way eventually attracting some people to join us ...
[ Now senseless idea about new structure deleted]
>> I target a code Sequence as follows:
>> volume vol;
>> record catalog;
>> btreee tree;
>> UInt32 cnid;
>> volume_open(&vol, "/dev/hdx", HFSP_READONLY);
>> record_open_catalog(&catalog, &vol);
>> btree_open(&tree, &catalog);
>> icnid = btree_find_recursive(&tree, "MyFile.txt");
>> btree_copy_data(&tree, "/home/user/MyFile.txt");
>>
>> This would be the core for a hfspcp tool ...
> It seems more logical to me to have the following:
> volume_open(&vol, device, flags);
> btree_open(&tree, &vol, CATALOG_CNID);
> record_init(&record, &tree);
> find_record(&record, searchkey);
> catrecord_copy_data(&record, outfile);
> This seems more in line with the actual relationships of the
> on disk structures, whereas I would find the other way to
> be somewhat misleading. In this case, the record would need
> the tree rather than the fork, and the tree needs the fork.
> However, I could just be biased by the way I wrote it. :)
I think you are right here. Now since I found my basic mistakes
I can go ahead and implement and test more utility functions.
I will restructuree it in a away that the record will be "below"
the btree.
Thanks for you comments, it really pushed me foreward to discuss
things with someone.
Greetings,
Klaus Halfmann aka "Hasi".
P.S : (You get my Libra Signature when I write to you from work ...)
