Re: [apacheds] JXplorer compatibility; some strangeness with names

[Marc Boorshtein <mb...@yahoo.com>]

Hey chris,

I was just curiose how far along your ldap backend is?

marc


--- Marc Boorshtein <mb...@yahoo.com> wrote:

> I've run into similar issues with jndi/java based
> ldap
> proxy.  The issue is generally because of
> conversions
> from UTF-8 -> UCS-2 -> UTF-8...especially when
> dealing
> with windows.  
> 
> As for the way jzplorer is displaying it, I've
> generally seen servers return UTF-8 characters in
> one
> of 2 ways:
> 1.  Escaped using a "\", so for isntance "\C5" 
> 2.  Assume the client is reading in the characters
> as
> UTF-8
> 
> I've never seen them encoded with a "%" before
> though...is that part of the RFC?
> 
> marc
> 
> 
> --- chris@pegacat.com wrote:
> > Hi Folks, 
> > 
> >    I've been messing around with custom
> partitions,
> > and have a couple of 
> > questions... 
> > 
> >   first (easy) question; is there any way to make
> > the sending of timestamp 
> > information optional, or should this be done
> inside
> > the partition?  At the 
> > moment timestamp info is sent as another
> attribute,
> > and the directory I'm 
> > using wants to set that stuff itself (as it
> doesn't
> > trust clients :-) )... 
> > 
> >   second (yucky) question: I'm having trouble with
> > the handling of 'special' 
> > characters in names.  They seem to get mangled
> > between my partition passing 
> > them back (as a naming enumeration of search
> result
> > objects), and apacheds 
> > making the ldap response. 
> > 
> >   Specifically, this is the raw LDAP straight from
> > my directory... 
> > 
> > 0000: 30 82 01 6F 02 01 0A 64   82 01 68 04 44 63
> 6E
> > 3D  0..o...d..h.Dcn=
> > 0010: 56 69 76 69 65 6E 6E 65   20 5C 22 4C 45 56
> 45
> > 52  Vivienne \"LEVER
> > 0020: 5C 22 2C 6F 75 3D 41 64   6D 69 6E 69 73 74
> 72
> > 61  \",ou=Administra
> > 0030: 74 69 6F 6E 2C 6F 75 3D   43 6F 72 70 6F 72
> 61
> > 74  tion,ou=Corporat
> > 0040: 65 2C 6F 3D 44 45 4D 4F   43 4F 52 50 2C 63
> 3D
> > 41  e,o=DEMOCORP,c=A
> > 0050: 55 30 82 01 1E 30 1E 04   0B 6F 62 6A 65 63
> 74
> > 43  U0...0...objectC
> > 0060: 6C 61 73 73 31 0F 04 0D   69 6E 65 74 4F 72
> 67
> > 50  lass1...inetOrgP
> > 0070: 65 72 73 6F 6E 30 18 04   02 63 6E 31 12 04
> 10
> > 56  erson0...cn1...V
> > 0080: 69 76 69 65 6E 6E 65 20   22 4C 45 56 45 52
> 22
> > 30  ivienne "LEVER"0
> > 0090: 0D 04 02 73 6E 31 07 04   05 4C 45 56 45 52
> 30
> > 24  ...sn1...LEVER0$
> > (etc.) 
> > 
> >   This is what I get from apacheDS... 
> > 
> > 0000: 30 82 01 42 02 01 09 64   82 01 3B 04 17 63
> 6E
> > 3D  0..B...d..;..cn=
> > 0010: 56 69 76 69 65 6E 6E 65   20 5C 5C 22 4C 45
> 56
> > 45  Vivienne \\"LEVE
> > 0020: 52 5C 5C 22 30 82 01 1E   30 0D 04 02 73 6E
> 31
> > 07  R\\"0...0...sn1.
> > 0030: 04 05 4C 45 56 45 52 30   1D 04 0F 74 65 6C
> 65
> > 70  ..LEVER0...telep
> > (etc.) 
> > 
> >   obviously it's shorter, and it misses the higher
> > RDNs (jndi just gives the 
> > relative name; my partition can expand it if
> > necessary?).  But more 
> > importantly, it messes up the special characters,
> > and JXplorer displays the 
> > entry as: Vivienne%20%5c%5c%22LEVER%5c%5c%22 . 
> > yikes.  A similar problem 
> > occurs with names that just have a space (e.g.
> 'fred
> > nurk'); the spaces get 
> > translated to %20, so it's not just the quote
> > characters above. 
> > 
> >   If I read back a similar entry from the ApacheDS
> > user partition, it all 
> > works pretty much as expected (e.g. similar to the
> > first option above). 
> > 
> > 0000: 30 30 02 01 0A 65 2B 0A   01 00 04 24 75 69
> 64
> > 3D  00...e+....$uid=
> > 0010: 66 72 65 64 20 5C 22 6E   75 72 6B 5C 22 2C
> 6F
> > 75  fred \"nurk\",ou
> > 0020: 3D 75 73 65 72 73 2C 6F   75 3D 73 79 73 74
> 65
> > 6D  =users,ou=system
> > 0030: 04 00                                       
>  
> >     .. 
> > 
> >   ... and the entry displays correctly in
> JXplorer. 
> > 
> >   So...
> >   1) Does a partition need to return full names
> > rather than relative names 
> > in the BackingStore 'search()' method?
> >   2) Do these names need to be pre-processed in
> any
> > way to avoid multiple 
> > escapings of characters in names? 
> > 
> >   Any thoughts? 
> > 
> >      - Chris 
> > 
> > 
> 

Re: [apacheds] ldap backed

[Chris Betts <ch...@pegacat.com>]

Hi Marc,

     I ended up writing a generic 'jndi backend', and when I set it to 
use the default sun jndi / ldap provider, it works very nicely as an 
ldap backend.  There's still the problem with special characters in the 
names being mangled, but I haven't had time to chase that down yet...

     - Chris


On 03/05/2005, at 11:42 AM, Marc Boorshtein wrote:

> Hey chris,
>
> I was just curiose how far along your ldap backend is?
>
> marc
>
>
> --- Marc Boorshtein <mb...@yahoo.com> wrote: