afpfs-ng-devel — Discussion about various technical issues with afpfs-ng.

Re: [Afpfs-ng-devel] GSS UAM support.

[Alex d. <ale...@gm...>]

Omri,

I don't know that anyone's working on a Kerberos or GSS UAM, although
Derrik Pates wrote all the non-trivial UAMs.  I'd be happy to
entertain those patches.

In having a quick scan at the netatalk GSS code, you'll need some of
the primitives created (eg. FPGetSessionToken).  Those are trivial for
me to write now, so if you let me know what you need, I can whip those
up pretty quickly.

I warn you that the code in CVS right now is in somewhat rough shape.

BTW, it is a lot easier to mail to this mailing list if you're
subscribed to the list.

- Alex

On 9/7/07, Omri Schwarz <ocs...@mi...> wrote:
> Hi again.
>
> I've installed 0.4.1 and will upgrade when the new tarball
> propagates throguh the SF mirrors. In the meantime I thought I'd
> ask, is anyone working on enabling the GSS UAM?
>
> Thanks again,
>
>
> Omri Schwarz --- ocs...@mi...
> Timeless wisdom of biomedical engineering: "Noise is principally
> due to the presence of the patient." -- R.F. Farr
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Afpfs-ng-devel mailing list
> Afp...@li...
> https://lists.sourceforge.net/lists/listinfo/afpfs-ng-devel
>

[Afpfs-ng-devel] GSS UAM support.

[Omri S. <ocs...@MI...>]

Hi again.

I've installed 0.4.1 and will upgrade when the new tarball
propagates throguh the SF mirrors. In the meantime I thought I'd
ask, is anyone working on enabling the GSS UAM?

Thanks again,


Omri Schwarz --- ocs...@mi...
Timeless wisdom of biomedical engineering: "Noise is principally
due to the presence of the patient." -- R.F. Farr

Re: [Afpfs-ng-devel] Borked tarball.

[Alex d. <ale...@gm...>]

Thanks for noticing, I've re-uploaded a fixed version.

- Alex

On 9/7/07, Omri Schwarz <ocs...@mi...> wrote:
> Hi, all.
>
> Just to note, the latest afpfs-ng tarball posted on
> Source forge will not bunzip.
>
>
> Omri Schwarz --- ocs...@mi...
> Timeless wisdom of biomedical engineering: "Noise is principally
> due to the presence of the patient." -- R.F. Farr
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Afpfs-ng-devel mailing list
> Afp...@li...
> https://lists.sourceforge.net/lists/listinfo/afpfs-ng-devel
>

[Afpfs-ng-devel] Borked tarball.

[Omri S. <ocs...@MI...>]

Hi, all.

Just to note, the latest afpfs-ng tarball posted on
Source forge will not bunzip.


Omri Schwarz --- ocs...@mi...
Timeless wisdom of biomedical engineering: "Noise is principally
due to the presence of the patient." -- R.F. Farr

[Afpfs-ng-devel] What's going into CVS

[Alex d. <ale...@gm...>]

Over the next couple of days, I'm going to be reworking the entire
layout of what's in CVS, and it will be compilable but unstable.

Of significance is the move to libafpclient, which will allow you have
different clients using the same protocol code.

Once the library is built, it will allow you to build any number of
clients, of which fuse is just one.

I am developing a simple command line version that shows this; this is
useful for transferring individual files rather than having to mount a
full filesystem.  This makes room for the gnome-vfs2 client...

- Alex

[Afpfs-ng-devel] Announcing afpfs-ng 0.4.3

[Alex d. <ale...@gm...>]

It may have taken 7 months, but I've released afpfs-ng 0.4.3.

You can download it directly from:

http://sourceforge.net/project/showfiles.php?group_id=179882&package_id=208206&release_id=537782

I'd be happy to upload Debian, Ubuntu, etc packages i people can provide them.


Here's the NEWS:


What is new in afpfs-ng-0.4.3, September 8, 2007

New features in this release include:

- UTF8 internationalization of filenames, volumes and servers, mostly
written by Michael Ulbrich

- proper uid/gid mapping to enable environments with a common or
different user directory

- DHX2 UAM from Derrik Pates

- fixes to deal with the Apple Airport Extreme quirks

- fixes to deal with netatalk quirks

- improved status output for debugging

- various small bug fixes (Paul Borman, Volker Grabsch)


This is the last release before a rework that introduces libafpclient to
handle multiple clients.

You can get afpfs-ng from http://afpfs-ng.sourceforge.net .

[Afpfs-ng-devel] Input/Output Error

[Novaoblivion <nov...@gm...>]

I am using afpfs-ng to connect to a AFP server running on my AppleTV.  
My Mac can connect to the share and copy files to and from it fine.  
However from my Ubuntu system using afpfs I am having some problems.  
I was able to get it to compile and install fine, I can mount a share  
from the command line and browse it. The problem comes when I try to  
copy something to the share. I get a Input/Output error both when  
trying from Nautilus or the command line. I am using version 0.4.1,  
on Ubuntu Feisty with Fuse 2.6.3. I saw another post with a similar  
problem on here but using 0.4.1 seemed to fix it for him. I am not  
sure what kind of information you need to track down the problem, but  
please let me know and I will get it for you. Thanks!

Re: [Afpfs-ng-devel] accented characters in UTF-8 filenames

[Michael U. <mu...@re...>]

Alex, all,

a first approach to UTF8 support for afpfs-ng has been committed to CVS. 
Currently only convert_utf8dec_to_utf8pre() for decomposed -> 
precomposed UTF8 conversion is actually implemented.

IMHO the corresponding convert_utf8pre_to_utf8dec() is not necessary 
since my tests have shown so far that the server decomposes all UTF8 
strings before writing them as AFPNames to the filesystem.

Please try it and let me know what you think!

Regards ... Michael

Alex deVries wrote:
> Michael,
> 
> Finally, I have rewritten some code to handle better filename handling
> so there's one simple point to add in such translation code.  It is
> now in CVS.
> 
> If you look in codepage.c, you'll see:
> 
> int convert_utf8dec_to_utf8pre(const char *src, int src_len,
>         char * dest, int dest_len);
> int convert_utf8pre_to_utf8dec(const char * src, int src_len,
>         char * dest, int dest_len);
> 
> There's some terrible code in there right now that will translate just
> one character, that's a test.  If you can fill those in, that would be
> terrific.
> 
> This code works for just filenames and volume names; usernames are coming next.
> 
> Let me know what you think...
> 
> - Alex

[Afpfs-ng-devel] afpfs-ng from CVS

[Michael U. <mu...@re...>]

Hi Alex,

I've encountered a small problem with the current CVS version of 
afpfs-ng. It shows up when trying to create some file on a mounted volume:

cp some.file /home/mul/mnt/Public/ABC

cp: cannot create regular file `/home/mul/mnt/Public/ABC/some.file': 
Input/output error

ls on the volume shows an empty file /home/mul/mnt/Public/ABC/some.file

A following second attempt of the same command is successful!

Connecting to a different AFP machine gives a slightly different result:

cp some.file /home/mul/mnt/some/where

cp: cannot create regular file `/home/mul/mnt/some/where': Permission denied

ls on the volume again shows an empty file.

A second attempt fails with the same error. On each attempt syslog puts 
out 'openfork response is too short'.

Other operations on the mounted volume have been without problems so far.

Any ideas would be greatly appreciated ... ;-)

Thanks + Best regards ... Michael

[Afpfs-ng-devel] BSD support?

[Alex d. <ale...@gm...>]

I've seen some NetBSD packages out there, which is interesting and
surprising.  Has anyone on the list had any success with it?  I'd
really like to make changes to the source code to get it to work for
BSD out of the box, and remove the "for Linux" notes in various
afpfs-ng descriptions.

- Alex

[Afpfs-ng-devel] The situation with afpfs-ng-cvs, the CVS mailing list

[Alex d. <ale...@gm...>]

I've notice some people subscribing to the afpfs-ng-cvs mailing list,
which is, uh, optimistic.

I haven't yet been able to setup things on SF so it'll actually
announce changes made to CVS.

However, there are some updates to CVS on the road to implementing
UTF8 support, and other changes will be trickling in.  When UTF8
support happens, we'll release 0.5.

- Alex

Re: [Afpfs-ng-devel] accented characters in UTF-8 filenames

[Alex d. <ale...@gm...>]

Michael,

Finally, I have rewritten some code to handle better filename handling
so there's one simple point to add in such translation code.  It is
now in CVS.

If you look in codepage.c, you'll see:

int convert_utf8dec_to_utf8pre(const char *src, int src_len,
        char * dest, int dest_len);
int convert_utf8pre_to_utf8dec(const char * src, int src_len,
        char * dest, int dest_len);

There's some terrible code in there right now that will translate just
one character, that's a test.  If you can fill those in, that would be
terrific.

This code works for just filenames and volume names; usernames are coming next.

Let me know what you think...

- Alex

On 6/11/07, Alex deVries <ale...@gm...> wrote:
> Michael,
>
> Hey, that's great!
>
> On 6/11/07, Michael Ulbrich <mu...@re...> wrote:
> > After checking copy_path() in utils.c I assume that it's not necessary
> > to do any conversion from UTF8 to UTF8-MAC in this routine. AFAI
> > understand copy_path() is called with client side path names to
> > transform them into AFP names. From what I've read so far the filesystem
> > on the MAC side will take care of converting precomposed to decomposed
> > unicode chararcters. Some experiments with file and directory names
> > containing "unusual" characters seem to support this assumption.
>
> Yes, that's correct.  I'll change the calling path a bit, but you have
> the right idea.
>
> >
> > The problem arises in the opposite direction: after reading AFP names in
> > UTF8-MAC these names should be converted to UTF8 before returning them
> > to the client side. I checked the code but must confess that I'm a
> > little lost here ... Currently I've tried to convert data contained in
> > the 'name' field of struct afp_file_info in afp_readdir() in fuse_int.c
> >    but without success so far. Any pointers would be greatly appreciated!
>
> Right.  I haven't put the right hooks in to handle the conversions to
> UTF8-MAC and from UTF8-MAC.  We should do this in a way that will work
> well with other encodings too, so that AFP 2.x conversions can be
> added at a later date.  I can do that some time this week.
>
> I'll clean up what's in CVS with those hooks so you have something to
> patch against.  Once that's all working, we'll spin 0.4.5 or similiar.
>
> I'm still working on a side project to setup arbitrary afpfsd clients,
> of which one is FUSE.  The second one is a command line tool (similiar
> to an ftp client), and there's room for a gnome-vfs client later.
>
> Danke!
>
> - Alex
>

Re: [Afpfs-ng-devel] afp_client running under RHELv5 connecting to OSX tiger server

[Alex d. <ale...@gm...>]

On 7/16/07, Lokke Highstein <lh...@co...> wrote:
> > 2. User-specific mounts
> >
> > Either in conjunction or in replacement to the above, you could get
> > each user to do their own mount on login.  This would require that the
> > user enter authentication information.
>
> so i would have to edit their .cshrc files to run the afp_client
> command on login?  that would mean that their passwords would be in
> clear text in the .cshrc file if i am following you.

In afpfs-ng 0.4.1, you can in fact have more than one user doing a
mount at a time.  It sounds like there's some confusion over this.  It
will just start up one afpfsd process per user.  This isn't a big
overhead.

In afpfs-ng 0.4.1, if you don't enter the '-p passwd' flag with
afp_client, it will prompt you for it.

> this would work if it was set up to function like automount under
> solaris, but i don't think it's set up like that.

Do you know how Solaris does the authentication for automount?

> yes.  ideally i would like it to function much in the way NFS does
> (without the annoying bugs and slowness)  i.e.  upon startup, the
> root user would automatically mount the exported directories at (for
> instance) /data/directory1  /data/directory2 /alternate_data/directory3
>
> all the users need the same 10 directories, and they need them
> mounted at the same places in the filesystem, if that makes sense.

Okay.  What authentication would root use to do this original mount?

Good luck,

- A

Re: [Afpfs-ng-devel] afp_client running under RHELv5 connecting to OSX tiger server

[Lokke H. <lh...@co...>]

On Jul 16, 2007, at 3:05 PM, Alex deVries wrote:

> Lokke,
>
> On 7/16/07, Lokke Highstein <lh...@co...> wrote:
>> i am trying to get rid of the solaris server (it is old and out of
>> it's service contract) and replace it with a red hat enterprise linux
>> server (v.5)
>
> You will greatly improve your karma with this transition. Fewer
> parking tickets, sunnier days, better sales, etc.

thanks!  i thought as much, but appreciate the confirmation.

>> one of the goals i have for this is to do away with NFS mounting, as
>> it is buggy under OSX, and i was hoping to get some sort of AFP mount
>> working, as it seems to be more reliable and seems to be faster than
>> NFS (at least on apple client machines here in the lab.)
>
> Yes, afpfs-ng should be able to help you with this.
>
> I think you have three choices for doing the global mount system
> you're suggesting:
>
> 1. Set up the mount as one user.
>
> This can actually work quite well if you're using read only data, or
> the data can be accessed using just one account.
>
> Afpfs-ng currently has a problem with username mapping between the
> client and server for some scenarios.
>
> This won't work if each user has private data, which is likely the  
> case.

unfortunately we have over 20 users and there has to be private data,  
at least on the group level.

> 2. User-specific mounts
>
> Either in conjunction or in replacement to the above, you could get
> each user to do their own mount on login.  This would require that the
> user enter authentication information.

so i would have to edit their .cshrc files to run the afp_client  
command on login?  that would mean that their passwords would be in  
clear text in the .cshrc file if i am following you.

this would work if it was set up to function like automount under  
solaris, but i don't think it's set up like that.

> 3. Provide patches
>
> afpfs-ng is a community project.  I can guide you as to what to change
> to address this relevant use case.

we are getting way out of my depth of knowledge, here.  i am hardly a  
developer.  in fact i'm not much of a sysadmin to be honest.

> In your ideal case, how would you like authentication to be handled?
> Would each user only have to login to the client, then their directory
> would magically appear?

yes.  ideally i would like it to function much in the way NFS does  
(without the annoying bugs and slowness)  i.e.  upon startup, the  
root user would automatically mount the exported directories at (for  
instance) /data/directory1  /data/directory2 /alternate_data/directory3

all the users need the same 10 directories, and they need them  
mounted at the same places in the filesystem, if that makes sense.

>> i have also been getting errors authenticating when trying to test
>> the afp_client application, but i don't want to spend too much time
>> on that if it turns out that it won't meet my needs for the moment.
>
> Let me know.  This is usually a problem of not having the right crypto
> libraries installed at configure time.

i'm not so worried about this, as long as i am right about the  
mounting method not being quite what i am looking for, i can't spend  
the time troubleshooting authentication issues.

> I love afpfs-ng, and I've done what I can with it.  As the
> documentation suggests, afpfs-ng is not tested for production work.
> Use it at your own risk.  I say this to anyone who wants to use it.

this is most likely the biggest problem for me, as i am in a  
production environment, and need to use stable software.  most likely  
i will continue to monitor the progress and will use it when it gets  
out of the beta stage.

> Until the kitchen reno's done, I can't spend the time I'd like to  
> on it.

understood.  priorities are important.  i need to renovate my kitchen  
as well!

thanks for the prompt reply.

lokke

Re: [Afpfs-ng-devel] afp_client running under RHELv5 connecting to OSX tiger server

[Alex d. <ale...@gm...>]

Lokke,

On 7/16/07, Lokke Highstein <lh...@co...> wrote:
> i am trying to get rid of the solaris server (it is old and out of
> it's service contract) and replace it with a red hat enterprise linux
> server (v.5)

You will greatly improve your karma with this transition. Fewer
parking tickets, sunnier days, better sales, etc.

> one of the goals i have for this is to do away with NFS mounting, as
> it is buggy under OSX, and i was hoping to get some sort of AFP mount
> working, as it seems to be more reliable and seems to be faster than
> NFS (at least on apple client machines here in the lab.)

Yes, afpfs-ng should be able to help you with this.

I think you have three choices for doing the global mount system
you're suggesting:

1. Set up the mount as one user.

This can actually work quite well if you're using read only data, or
the data can be accessed using just one account.

Afpfs-ng currently has a problem with username mapping between the
client and server for some scenarios.

This won't work if each user has private data, which is likely the case.

2. User-specific mounts

Either in conjunction or in replacement to the above, you could get
each user to do their own mount on login.  This would require that the
user enter authentication information.

3. Provide patches

afpfs-ng is a community project.  I can guide you as to what to change
to address this relevant use case.


In your ideal case, how would you like authentication to be handled?
Would each user only have to login to the client, then their directory
would magically appear?

>
> i have also been getting errors authenticating when trying to test
> the afp_client application, but i don't want to spend too much time
> on that if it turns out that it won't meet my needs for the moment.

Let me know.  This is usually a problem of not having the right crypto
libraries installed at configure time.

I love afpfs-ng, and I've done what I can with it.  As the
documentation suggests, afpfs-ng is not tested for production work.
Use it at your own risk.  I say this to anyone who wants to use it.

Until the kitchen reno's done, I can't spend the time I'd like to on it.

- Alex

[Afpfs-ng-devel] afp_client running under RHELv5 connecting to OSX tiger server

[Lokke H. <lh...@co...>]

hello,

i am new to this list and to afpfs-ng

i currently administer a lab where we are using a solaris server to  
run analysis of experimental data.

we are using an OSX tiger server as a fileserver, with Xraid as  
storage.  the solaris server uses NFS to mount the exported  
directories from the OSX file server.

we are using another OSX tiger server as an open directory master  
(open LDAP) and a dns server.

i am trying to get rid of the solaris server (it is old and out of  
it's service contract) and replace it with a red hat enterprise linux  
server (v.5)

one of the goals i have for this is to do away with NFS mounting, as  
it is buggy under OSX, and i was hoping to get some sort of AFP mount  
working, as it seems to be more reliable and seems to be faster than  
NFS (at least on apple client machines here in the lab.)

ideally i would be able to AFP mount various directories at certain  
points in the file system on the linux server, and have these shares  
available to all users.

from what i have read so far, that won't be possible, as each user  
has to mount their own AFP mounts, is that correct?

i have also been getting errors authenticating when trying to test  
the afp_client application, but i don't want to spend too much time  
on that if it turns out that it won't meet my needs for the moment.

if there is a way to globally mount AFP shares, or even automatically  
mount the shares as users log in, that will be fine.  otherwise i'll  
stick with NFS for the moment.

thanks for any help or advice on this

-lokke

Re: [Afpfs-ng-devel] Problems compiling afpfs-ng

[James D. <ja...@mi...>]

Thanks for that - seems to have done the trick.

james

On Fri, 29 Jun 2007, Michael Ulbrich wrote:

> James Dyer wrote:
> > I'm having problems getting afpfs-ng to compile; can anyone help me get 
> > this working?
> > 
> > The configure stage appeared to complete OK, but when I ran make, it 
> > errors out with:
> > 
> > if gcc -DHAVE_CONFIG_H -I. -I. -I.    -g -Wall -D_FILE_OFFSET_BITS=64  -g 
> > -O2 -MT afpfsd-fuse_int.o -MD -MP -MF ".deps/afpfsd-fuse_int.Tpo" -c -o 
> > afpfsd-fuse_int.o `test -f 'fuse_int.c' || echo './'`fuse_int.c; \
> > then mv -f ".deps/afpfsd-fuse_int.Tpo" ".deps/afpfsd-fuse_int.Po"; else rm 
> > -f ".deps/afpfsd-fuse_int.Tpo"; exit 1; fi
> > In file included from fuse_int.c:24:
> > /usr/include/asm/fcntl.h:70: error: redefinition of `struct flock'
> > fuse_int.c: In function `afp_getattr':
> > fuse_int.c:206: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_rmdir':
> > fuse_int.c:434: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_unlink':
> > fuse_int.c:494: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_readdir':
> > fuse_int.c:560: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_mknod':
> > fuse_int.c:678: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_release':
> > fuse_int.c:748: warning: cast to pointer from integer of different size
> > fuse_int.c:767: warning: cast to pointer from integer of different size
> > fuse_int.c:796: warning: cast to pointer from integer of different size
> > fuse_int.c: In function `afp_open':
> > fuse_int.c:823: warning: assignment makes integer from pointer without a 
> > cast
> > fuse_int.c:837: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_write':
> > fuse_int.c:1065: warning: cast to pointer from integer of different size
> > fuse_int.c:1086: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c:1090: warning: implicit declaration of function `volinfo_write'
> > fuse_int.c: In function `afp_read':
> > fuse_int.c:1282: warning: cast to pointer from integer of different size
> > fuse_int.c:1290: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c: In function `afp_truncate':
> > fuse_int.c:1466: warning: passing arg 2 of `apple_translate' discards 
> > qualifiers from pointer target type
> > fuse_int.c:1478: warning: cast to pointer from integer of different size
> > make[1]: *** [afpfsd-fuse_int.o] Error 1
> > make[1]: Leaving directory `/home/jad/afpfs-ng-0.4.1'
> > make: *** [all] Error 2
> > 
> > 
> > 
> > Snipped all of the configure stuff out, and most of the make stuff - let 
> > me know if it'd be of any use.
> > 
> > Running 2.6.9-1.667smp - FC3
> > 
> > Any and all help appreciated, as I think afpfs-ng is going to dig me out 
> > of a hole soon (assuming I can get it to compile)
> > 
> > Thanks,
> > 
> > j
> 
> Hi,
> 
> I have encountered this problem as well.
> 
> To get afpfs-ng compiled, I've temporarily commented the offending 
> definitions in /usr/include/asm/fcntl.h. Not a real "solution" but it 
> works ;-)
> 
> Hope that helps ... Michael
> 
> 
> 

-- 
July 27th, 2007 - System Administrator Appreciation Day - http://www.sysadminday.com/

Re: [Afpfs-ng-devel] Problems compiling afpfs-ng

[Michael U. <mu...@re...>]

James Dyer wrote:
> I'm having problems getting afpfs-ng to compile; can anyone help me get 
> this working?
> 
> The configure stage appeared to complete OK, but when I ran make, it 
> errors out with:
> 
> if gcc -DHAVE_CONFIG_H -I. -I. -I.    -g -Wall -D_FILE_OFFSET_BITS=64  -g 
> -O2 -MT afpfsd-fuse_int.o -MD -MP -MF ".deps/afpfsd-fuse_int.Tpo" -c -o 
> afpfsd-fuse_int.o `test -f 'fuse_int.c' || echo './'`fuse_int.c; \
> then mv -f ".deps/afpfsd-fuse_int.Tpo" ".deps/afpfsd-fuse_int.Po"; else rm 
> -f ".deps/afpfsd-fuse_int.Tpo"; exit 1; fi
> In file included from fuse_int.c:24:
> /usr/include/asm/fcntl.h:70: error: redefinition of `struct flock'
> fuse_int.c: In function `afp_getattr':
> fuse_int.c:206: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_rmdir':
> fuse_int.c:434: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_unlink':
> fuse_int.c:494: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_readdir':
> fuse_int.c:560: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_mknod':
> fuse_int.c:678: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_release':
> fuse_int.c:748: warning: cast to pointer from integer of different size
> fuse_int.c:767: warning: cast to pointer from integer of different size
> fuse_int.c:796: warning: cast to pointer from integer of different size
> fuse_int.c: In function `afp_open':
> fuse_int.c:823: warning: assignment makes integer from pointer without a 
> cast
> fuse_int.c:837: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_write':
> fuse_int.c:1065: warning: cast to pointer from integer of different size
> fuse_int.c:1086: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c:1090: warning: implicit declaration of function `volinfo_write'
> fuse_int.c: In function `afp_read':
> fuse_int.c:1282: warning: cast to pointer from integer of different size
> fuse_int.c:1290: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c: In function `afp_truncate':
> fuse_int.c:1466: warning: passing arg 2 of `apple_translate' discards 
> qualifiers from pointer target type
> fuse_int.c:1478: warning: cast to pointer from integer of different size
> make[1]: *** [afpfsd-fuse_int.o] Error 1
> make[1]: Leaving directory `/home/jad/afpfs-ng-0.4.1'
> make: *** [all] Error 2
> 
> 
> 
> Snipped all of the configure stuff out, and most of the make stuff - let 
> me know if it'd be of any use.
> 
> Running 2.6.9-1.667smp - FC3
> 
> Any and all help appreciated, as I think afpfs-ng is going to dig me out 
> of a hole soon (assuming I can get it to compile)
> 
> Thanks,
> 
> j

Hi,

I have encountered this problem as well.

To get afpfs-ng compiled, I've temporarily commented the offending 
definitions in /usr/include/asm/fcntl.h. Not a real "solution" but it 
works ;-)

Hope that helps ... Michael

[Afpfs-ng-devel] Problems compiling afpfs-ng

[James D. <ja...@mi...>]

I'm having problems getting afpfs-ng to compile; can anyone help me get 
this working?

The configure stage appeared to complete OK, but when I ran make, it 
errors out with:

if gcc -DHAVE_CONFIG_H -I. -I. -I.    -g -Wall -D_FILE_OFFSET_BITS=64  -g 
-O2 -MT afpfsd-fuse_int.o -MD -MP -MF ".deps/afpfsd-fuse_int.Tpo" -c -o 
afpfsd-fuse_int.o `test -f 'fuse_int.c' || echo './'`fuse_int.c; \
then mv -f ".deps/afpfsd-fuse_int.Tpo" ".deps/afpfsd-fuse_int.Po"; else rm 
-f ".deps/afpfsd-fuse_int.Tpo"; exit 1; fi
In file included from fuse_int.c:24:
/usr/include/asm/fcntl.h:70: error: redefinition of `struct flock'
fuse_int.c: In function `afp_getattr':
fuse_int.c:206: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_rmdir':
fuse_int.c:434: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_unlink':
fuse_int.c:494: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_readdir':
fuse_int.c:560: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_mknod':
fuse_int.c:678: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_release':
fuse_int.c:748: warning: cast to pointer from integer of different size
fuse_int.c:767: warning: cast to pointer from integer of different size
fuse_int.c:796: warning: cast to pointer from integer of different size
fuse_int.c: In function `afp_open':
fuse_int.c:823: warning: assignment makes integer from pointer without a 
cast
fuse_int.c:837: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_write':
fuse_int.c:1065: warning: cast to pointer from integer of different size
fuse_int.c:1086: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c:1090: warning: implicit declaration of function `volinfo_write'
fuse_int.c: In function `afp_read':
fuse_int.c:1282: warning: cast to pointer from integer of different size
fuse_int.c:1290: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c: In function `afp_truncate':
fuse_int.c:1466: warning: passing arg 2 of `apple_translate' discards 
qualifiers from pointer target type
fuse_int.c:1478: warning: cast to pointer from integer of different size
make[1]: *** [afpfsd-fuse_int.o] Error 1
make[1]: Leaving directory `/home/jad/afpfs-ng-0.4.1'
make: *** [all] Error 2



Snipped all of the configure stuff out, and most of the make stuff - let 
me know if it'd be of any use.

Running 2.6.9-1.667smp - FC3

Any and all help appreciated, as I think afpfs-ng is going to dig me out 
of a hole soon (assuming I can get it to compile)

Thanks,

j

-- 
July 27th, 2007 - System Administrator Appreciation Day - http://www.sysadminday.com/

Re: [Afpfs-ng-devel] accented characters in UTF-8 filenames

[Alex d. <ale...@gm...>]

Michael,

Hey, that's great!

On 6/11/07, Michael Ulbrich <mu...@re...> wrote:
> After checking copy_path() in utils.c I assume that it's not necessary
> to do any conversion from UTF8 to UTF8-MAC in this routine. AFAI
> understand copy_path() is called with client side path names to
> transform them into AFP names. From what I've read so far the filesystem
> on the MAC side will take care of converting precomposed to decomposed
> unicode chararcters. Some experiments with file and directory names
> containing "unusual" characters seem to support this assumption.

Yes, that's correct.  I'll change the calling path a bit, but you have
the right idea.

>
> The problem arises in the opposite direction: after reading AFP names in
> UTF8-MAC these names should be converted to UTF8 before returning them
> to the client side. I checked the code but must confess that I'm a
> little lost here ... Currently I've tried to convert data contained in
> the 'name' field of struct afp_file_info in afp_readdir() in fuse_int.c
>    but without success so far. Any pointers would be greatly appreciated!

Right.  I haven't put the right hooks in to handle the conversions to
UTF8-MAC and from UTF8-MAC.  We should do this in a way that will work
well with other encodings too, so that AFP 2.x conversions can be
added at a later date.  I can do that some time this week.

I'll clean up what's in CVS with those hooks so you have something to
patch against.  Once that's all working, we'll spin 0.4.5 or similiar.

I'm still working on a side project to setup arbitrary afpfsd clients,
of which one is FUSE.  The second one is a command line tool (similiar
to an ftp client), and there's room for a gnome-vfs client later.

Danke!

- Alex

Re: [Afpfs-ng-devel] accented characters in UTF-8 filenames

[Michael U. <mu...@re...>]

Hi Alex,

I have some pieces of code hacked together to do the necessary 
conversions from UTF8-MAC (decomposed) to UTF8 (precomposed).

After checking copy_path() in utils.c I assume that it's not necessary 
to do any conversion from UTF8 to UTF8-MAC in this routine. AFAI 
understand copy_path() is called with client side path names to 
transform them into AFP names. From what I've read so far the filesystem 
on the MAC side will take care of converting precomposed to decomposed 
unicode chararcters. Some experiments with file and directory names 
containing "unusual" characters seem to support this assumption.

The problem arises in the opposite direction: after reading AFP names in 
UTF8-MAC these names should be converted to UTF8 before returning them 
to the client side. I checked the code but must confess that I'm a 
little lost here ... Currently I've tried to convert data contained in 
the 'name' field of struct afp_file_info in afp_readdir() in fuse_int.c 
   but without success so far. Any pointers would be greatly appreciated!

It's funny (and confusing from time to time ;-) : Doing a 'ls' in a 
xterm on a mounted fs with "funny" names and redirecting the output to a 
file will show decomposed unicode as the file's contents. The displayed 
  file names in the terminal window are correct. Transferring these 
names by copy+paste from the terminal window to a file will result in 
precomposed unicode ...

Best regards ... Michael

Alex deVries wrote:
> Michael,
> 
> You're absolutely right.  There need to be conversions between
> precomposed and decomposed UTF8 filenames.  Just because my native
> language is English doesn't mean that we shouldn't have full
> internationalization support in afpfs-ng!
> 
> I'm never proud of my code, and a particularly hacky bit is in
> copy_path() in utils.c.  There, you can see that I'm hardcoding the
> unicode hint, etc.  I'm sure this isn't correct, it just happens to
> work for non-extended characters.  This code is only ever hit for
> pathnames, not volume names.
> 
> The thing that would help all of this is having a snippet of C code
> that handle translation between pre and decomposed UTF8.  That's the
> part I haven't figured out yet.
> 
> I'll read the link below when I'm not at work.
> 
> You may have noticed a bit of a lag in afpfs-ng releases; I've been
> putting most of my energy into redesigning some of the internals for
> multiple clients.  This would allow a command line version (similiar
> to an FTP client), a gnome-vfs2 interface, etc to all talk to the same
> protocol handler.  The FUSE interface stays, also.
> 
> - Alex

Re: [Afpfs-ng-devel] afpfsd does not start

[Alex d. <ale...@gm...>]

Sean,

That's a bit odd, I wonder if it is a problem with PPC or BE.

Some questions:

1. What's the output when you run:
killall -9 afpfsd
/path/to/afpfsd -d

(specify the full path to afpfsd)

2. Do you see this on 0.4.2?

3. Do you have other versions of afpfs-ng installed somewhere on your rootfs?

- Alex

On 5/19/07, Sean Zimmermann <sv...@pi...> wrote:
> Hello. I recently compiled afpfs-ng 0.4.1 on my powerpc running Debian testing. When I launched the afp_client, I got the following:
>
> The afpfs daemon does not appear to be running, let me start it for you
> Usage: afpfsd [OPTION]
>   -l, --logmethod    Either 'syslog' or 'stdout'  -f, --foreground   Do not fork  -d, --debug        Does not fork, logs to stdout
> Version 0.4
> The afpfs daemon does not appear to be running, let me start it for you
> Usage: afpfsd [OPTION]
>   -l, --logmethod    Either 'syslog' or 'stdout'  -f, --foreground   Do not fork  -d, --debug        Does not fork, logs to stdout
> Version 0.4
> Trying to startup afpfsd: No such file or directory
>
> When I tried to run afpfsd on its own, I got the same message.
>
> Is there anything I can do to fix this?
>
> Sean
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Afpfs-ng-devel mailing list
> Afp...@li...
> https://lists.sourceforge.net/lists/listinfo/afpfs-ng-devel
>

[Afpfs-ng-devel] afpfsd does not start

[Sean Z. <sv...@pi...>]

Hello. I recently compiled afpfs-ng 0.4.1 on my powerpc running Debian testing. When I launched the afp_client, I got the following:

The afpfs daemon does not appear to be running, let me start it for you
Usage: afpfsd [OPTION]
  -l, --logmethod    Either 'syslog' or 'stdout'  -f, --foreground   Do not fork  -d, --debug        Does not fork, logs to stdout
Version 0.4
The afpfs daemon does not appear to be running, let me start it for you
Usage: afpfsd [OPTION]
  -l, --logmethod    Either 'syslog' or 'stdout'  -f, --foreground   Do not fork  -d, --debug        Does not fork, logs to stdout
Version 0.4
Trying to startup afpfsd: No such file or directory

When I tried to run afpfsd on its own, I got the same message.

Is there anything I can do to fix this?

Sean

Re: [Afpfs-ng-devel] Small compile bug in CVS

[Alex d. <ale...@gm...>]

Rob,

Thanks for the patch, I applied it somewhat blindly.

I am surprised at any success you've had with what's in CVS as it is
really meant to be very rough.  This weekend, I'll dump in all the
work I've been doing for the last few months and I am 100% sure that
it will break everything.  But I'm closer to a command-line version of
the client.

- Alex

On 5/18/07, Robin Perkins <rob...@in...> wrote:
> Hi List,
>
> Straight out of CVS afpfs-ng doesn't compile. Here is a tiny patch
> that I made to fix this. It should be fairly trivial.
>
> Thanks,
>
> Rob
>
> (Hopefully this works - After all I did use afpfs-ng to transfer the
> diff to my mac ! :)
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Afpfs-ng-devel mailing list
> Afp...@li...
> https://lists.sourceforge.net/lists/listinfo/afpfs-ng-devel
>
>
>

[Afpfs-ng-devel] Small compile bug in CVS

[Robin P. <rob...@in...>]

Hi List,

Straight out of CVS afpfs-ng doesn't compile. Here is a tiny patch  
that I made to fix this. It should be fairly trivial.

Thanks,

Rob

(Hopefully this works - After all I did use afpfs-ng to transfer the  
diff to my mac ! :)