Difference between revisions of "Notes20100607"

From Net-SNMP Wiki
Jump to: navigation, search
Line 8: Line 8:
 
* Releases
 
* Releases
 
** 5.2.x -- dts12
 
** 5.2.x -- dts12
*** '''is out!'''
+
*** '''5.2.6 is out!'''
 
** 5.4.x -- dts12
 
** 5.4.x -- dts12
*** '''is out!'''
+
*** '''5.4.3 is out!'''
 
** 5.6 -- hardaker
 
** 5.6 -- hardaker
 
*** '''pre1 this week'''
 
*** '''pre1 this week'''

Revision as of 20:38, 7 June 2010

Previous: 2010-05-17
Next: 2010-07-19

This is the meeting agenda/notes for the Meeting held on 2010-06-07.

Date and Time: 2010-06-07 20:00 UTC

Agenda

  • Next Meeting Date
    • July 19th
  • Releases
    • 5.2.x -- dts12
      • 5.2.6 is out!
    • 5.4.x -- dts12
      • 5.4.3 is out!
    • 5.6 -- hardaker
      • pre1 this week
    • Release Testing
      • hardaker hasn't made progress on the results mailer yet
      • It would be a good idea to mention the output mailer in the pre1 release announcement
      • Wiki page would be a good idea: Test Results with subpages per release
      • hardaker has a reference to a build/package system he needs to look into that may be like the old SF build system that went away
    • Binary Releases
  • Administrative
  • Technical Discussions
    • Test infrastructure -- hardaker
    • Whether or not the triple build infrastructure should be unified to a single build infrastructure.
      • EG, CMake
      • hardaker has a plan to extract the module changing and header generation functionality to an external script so that people can just run the header generation scripts without running full configure
        • Ideally it should be called from configure, but it could also be inline-sourced in both locations
    • Net-SNMP V5 get/set request handler documentation [1].
      • Real issues:
        1. man pages and online copies are often out of date
          • hardaker will try to add more automation to the release generation script near 5.6 publication
        2. no high-level tutorial about how the internal of the agent works
        3. tutorials need more data
        4. many man pages are too small or useless
          • Dave will come up with a list of potential man page changes/cut-list and post to -coders
    • Perl
      • NetSNMP::agent regression test 5 fails [2].
      • Status of CPAN (5.4.1)
        • Wes explains the history:
[12:40] <hardaker> so I'll start with history
[12:40] <hardaker> Way back in the day when Giovanni Marzot wrote the original perl binding he put it in CPAN.
[12:41] <hardaker> Then I met him and we had a good long conversation about how much of a pain C-bindings are in CPAN
[12:41] <hardaker> we agreed that we'd move the core source to Net-SNMP itself so that it could be more easily aligned.
[12:41] <hardaker> that's all been well and dandy.
[12:41] <hardaker> then we started updating CPAN again after most-major-version-updated
[12:41] <hardaker> at which point a few people started screaming.
[12:42] <hardaker> the versions are tightly bound (and there is a version check to ensure that you compile the right perl module against the  right Net-SNMP source )
[12:42] <hardaker> so...  if CPAN has 5.6, for example, but you have 5.5.3 on your system you'll get:
[12:42] <hardaker> 1) an update suggestion from CPAN if you use CPAN to manage your verisons
[12:43] <hardaker> 2) a make soft-error saying you have a version mismatch if you try to let CPAN update it
[12:43] <hardaker> Now, someone has complained that the CPAN source is too old.
[12:43] <hardaker> I don't have a solution.
[12:43] <hardaker> there is no version to put in CPAN that will leave people happy.
[12:43] <hardaker> so suggestions welcome about what version we should publish to CPAN.
        • No resolution today; many different versions have many different Net-SNMP source bundles
    • jsafranek: hrFSTable and hrStorageTable persistent indexes - partially implemented in new host/hrh_filesys.c and hardware/fsys/*, but I miss some pieces :(
      • Done! Thanks to dts12
    • Cygwin --enable-shared build [3]: only possible if the circular dependency between libnetsnmpagent and libnetsnmphelpers is resolved -- see also [4]. What is the status of this last issue ?
      • done!
    • "Zero compiler warnings" policy for 5.6+?
      • Thanks especially to Bart, we're down to very few compiler warnings in trunk. Shall we adopt a "Zero compiler warnings" policy from now onwards (not applying to 5.5.x and earlier), meaning that newly introduced warnings need to be fixed ASAP?
      • Nice in theory, but very difficult in practice to do architecture/compiler, etc, differences
  • Round Table -- What are ya up to?
    • hardaker: (D)TLS and testing
    • magfr: settable paths (The ability to set the search paths in the generated binaries independently of the prefix.)
    • dts12: investigate inline documentation