Re: [HACKERS] LibreOffice driver 2: MIT Kerberos vs Microsoft Kerberos

2011-12-16 Thread Lionel Elie Mamane
On Tue, Dec 13, 2011 at 05:09:01PM -0500, Stephen Frost wrote: * Greg Smith (g...@2ndquadrant.com) wrote: Given that pgAdmin III has given up on MIT KRB5, would you feel doing the same is appropriate for LibreOffice too? Yes, I'd encourage LibreOffice to drop MIT Kerberos for Windows from

[HACKERS] LibreOffice driver 2: MIT Kerberos vs Microsoft Kerberos

2011-12-13 Thread Lionel Elie Mamane
(See part 1 for general introduction) LibreOffice currently - when it rebuilds libpq, such as for our official MS Windows binaries - links libpq against only the MS Windows built-in SSPI stuff, which if I understand well is an embrace and extend Kerberos 5 implementation. I wanted to understand

Re: [HACKERS] LibreOffice driver 2: MIT Kerberos vs Microsoft Kerberos

2011-12-13 Thread Stephen Frost
* Lionel Elie Mamane (lio...@mamane.lu) wrote: The gsslib parameter in the connection string won't work, but will that keep users from authenticating to some Kerberos domains, and/or are there other (interoperability?) issues that make it strongly desirable to link libpq with *both* SSPI *and*

Re: [HACKERS] LibreOffice driver 2: MIT Kerberos vs Microsoft Kerberos

2011-12-13 Thread Greg Smith
On 12/13/2011 09:18 AM, Stephen Frost wrote: The gist of the limitation is this- if you need to support decent encryption in a cross-realm environment on Windows XP-age systems, you need MIT KRB5. If you're on Windows 7 or something else recent, the built-in Windows stuff w/ AES works fine.

Re: [HACKERS] LibreOffice driver 2: MIT Kerberos vs Microsoft Kerberos

2011-12-13 Thread Stephen Frost
* Greg Smith (g...@2ndquadrant.com) wrote: This answers Lionel's question, but I'm curious for a more user impact opinion from you. Given that pgAdmin III has given up on MIT KRB5, would you feel doing the same is appropriate for LibreOffice too? It sounds like they really shouldn't take on