Difference between revisions of "Release Meeting Minutes/2010-07-20"
From K5Wiki
(New page: Zhanna Tsitkova, Thomas Hardjono, Greg Hudson, Tom Yu, Will Fiveash, Simo Sorce Plugin architecture -- "factory" versus concrete constructor? Are plugin handles domain-specific? ;Will: d...) |
|||
| Line 14: | Line 14: | ||
;Greg: OpenSSL conflict was a global initializer issue, so not really an autodiscovery problem. |
;Greg: OpenSSL conflict was a global initializer issue, so not really an autodiscovery problem. |
||
| − | Will says something about keeping stuff out of user- |
+ | Will says something about keeping stuff out of user-editable configs. |
Revision as of 12:24, 4 August 2010
Zhanna Tsitkova, Thomas Hardjono, Greg Hudson, Tom Yu, Will Fiveash, Simo Sorce
Plugin architecture -- "factory" versus concrete constructor? Are plugin handles domain-specific?
- Will
- don't use "void *"
- Greg
- Discovery for dynamic plugins? Current stuff works for Debian, but not for Red Hat. Consensus on krbdev list for using config fragments. OK for Simo?
- Simo
- yes
- Zhanna
- krbdev list prefers deferred loading. Want to give admins option to load everything at startup.
- Greg
- probably not a benefit for our use cases.
- Will
- agree with Greg.
- Zhanna
- Our experiences at Novell -- want to laod everything up front.
- Will
- For the DAL, when LDAP was introduced, DB2 was enabled by default, but not LDAP.
- Greg
- Is autodiscovery a problem for Will?
- Will
- OpenSSL dependency of PKINIT caused conflicts... most people using Solaris are not using smart cards.
- Greg
- OpenSSL conflict was a global initializer issue, so not really an autodiscovery problem.
Will says something about keeping stuff out of user-editable configs.
