Sophie

Sophie

distrib > Mageia > 4 > x86_64 > by-pkgid > b23c198eef6044fe9bb32a0acc43a274 > files > 44

hfsplusutils-1.0.4-10.mga4.x86_64.rpm

Hello Brad,

Yesterday I submitted the first "real" command to the CVS: hpmount.
You must do a make -f Makefile.cvs to get it running (Understanding
Well, it is not very usefull but the first step. My current plans
look like this:
- create hpls
- create hpcd
- create hpcp
- create hpunmount
? did I miss a reasonalbe, readonly command ?
- go on working on the extends overflow stuff

Ill try to push Suse, that they do a weekly build on the stuff
(So that I dont forget to add/commit a file) and supply the
linuxppc community with the tools. When I feel fine about the
tools on ppc we must find some fools^H^H^H^H^H nice people
that try our code on x86 and other platforms, too.
(Can you supply a http/ftp acces to some images ?)

When the readonly part is working, theres still is a lot to do:

- Need to write the supposed minimum checks to avoid duplicate
  CNIDs. Since this includes a complete tree traversal, so I can
  check for the btrees consistency, too.

- Need to write code to allocate and deallocate blocks in forks and 
  care about the allocation bitmaps. (append / truncate)

- Need to write code for allocating and deallocating nodes in the
  node allocation bitmaps.

I think I must write a node cache, to support atomic transactions
and to avoid the overhead of reading two blocks I currently have.

What I fear most is the code to rebalance the tree when
it is completly filled and all the other cases. Maybe Ill
tweak some images with MacOS to find out how its done there.

/me thinks a while about btree reorganisation and shudders

Uhm, I think I should write my Ideas into the documentation
first, perhaps you can point met to some code / mailing lists
I might study. I'll try to define some calculateable values 
that require a reorganisation.

/me trys to think even harder

Mhh, foget it, Ill take a pragmatically aproach: Ill do my
best in the simple cases and bail out if cant cope with
them right now (So most operations will succeed). After the
structure is ok Ill think about writing those (recursive) 
reorganisation fucntions.

Greetings,
	Klaus Halfmann,	aka Hasi