Difference between revisions of "Notes20110523"

From Net-SNMP Wiki
Jump to: navigation, search
 
(9 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
<!-- Reminder for Fall 2011 - will need to use "time=20" to allow for DST -->
 
<!-- Reminder for Fall 2011 - will need to use "time=20" to allow for DST -->
{{meetingtime|time=19|year=2011|month=05|day=23|nextyear=2011|nextmonth=06|nextday=??|prevyear=2011|prevmonth=04|prevday=25}}
+
{{meetingtime|time=19|year=2011|month=05|day=23|nextyear=2011|nextmonth=06|nextday=20|prevyear=2011|prevmonth=04|prevday=25}}
  
  
Line 6: Line 6:
  
 
* Next Meeting
 
* Next Meeting
 +
** June 20th
 
* Releases
 
* Releases
 
** 5.3.4
 
** 5.3.4
 +
*** '''rc1 at the end of the week'''
 
** 5.4.4
 
** 5.4.4
 +
*** '''rc1 at the end of the week'''
 
** 5.7
 
** 5.7
 +
*** '''pre2 out todayish'''
 
* Administrative
 
* Administrative
 
** Switching to GIT
 
** Switching to GIT
Line 16: Line 20:
 
***# switch everything but the 5.{3,4}.4 branches this week, and apply patches from the svn tree later to the git tree after releases are finished.
 
***# switch everything but the 5.{3,4}.4 branches this week, and apply patches from the svn tree later to the git tree after releases are finished.
 
***# delay the switch till after 5.{3,4}.4 final (as long as it's before 5.7).
 
***# delay the switch till after 5.{3,4}.4 final (as long as it's before 5.7).
 +
***#* '''Daves' preference and Wes is fine with it, so we'll go with this choice unless 5.{3,4}.4 drag out too long.
 
** Bisectability of the SVN/git tree.
 
** Bisectability of the SVN/git tree.
 +
*** '''Agreed: We shouldn't push broken commits because it means ''git bisect'' breaks if you can't test the broken code in the middle of a bisect'''
 +
** standardizing "please write to the list" notes? Possibly with a new wiki page with a longer explanation?
 +
*** Wes wrote one: [[Please write responses to the mailing lists]]
 +
**** feedback and editing appreciated
 +
**** start pointing to it?
 +
** LTS Timing
 +
*** From last time: 5.4 and 5.7 will be LTS
 +
*** How long is Long?
 +
*** '''Dave's proposal: keep 2 LTS branches active, ensuring at least 5 years.  Keep 3 branches otherwise'''
 +
*** Examples:
 +
**** 5.4L, 5.5, 5.6, 5.7L
 +
**** 5.4L, 5.6, 5.7L, 5.8
 +
**** 5.4L, 5.7L, 5.8, 5.9
 +
**** 5.7L, 5.8, 5.9, 5.10L
 
* Technical
 
* Technical
 +
** the infamous oid typedef issues
 +
*** '''5.6.1.1 still scheduled to go out'''
 +
*** '''Should we consider going to a 6.0, where we change all types to match (the biggest becoming all the u_longs used everywhere for ints).'''
 +
**** '''Wes notes that it may be possible to have a injectable handle that swaps types in the middle of a handler chain for legacy code'''
 +
*** '''general discussion tabled for future meetings about 6.0'''
 
** Windows Builds
 
** Windows Builds
 
*** Dave hopefully has windows working?  Needs wiki post about doing it.
 
*** Dave hopefully has windows working?  Needs wiki post about doing it.
 
* Round Table
 
* Round Table
 +
** dts12: 5.x.4
 +
** hardaker: 5.7 and git
 +
** bvanassche: wiki pages about git

Latest revision as of 20:07, 23 May 2011

Previous: 2011-04-25
Next: 2011-06-20

This is the meeting agenda/notes for the Meeting held on 2011-05-23.

Date and Time: 2011-05-23 19:00 UTC


Agenda

  • Next Meeting
    • June 20th
  • Releases
    • 5.3.4
      • rc1 at the end of the week
    • 5.4.4
      • rc1 at the end of the week
    • 5.7
      • pre2 out todayish
  • Administrative
    • Switching to GIT
      • Timing choices:
        1. switch now (ie, this week) for everything
        2. switch everything but the 5.{3,4}.4 branches this week, and apply patches from the svn tree later to the git tree after releases are finished.
        3. delay the switch till after 5.{3,4}.4 final (as long as it's before 5.7).
          • Daves' preference and Wes is fine with it, so we'll go with this choice unless 5.{3,4}.4 drag out too long.
    • Bisectability of the SVN/git tree.
      • Agreed: We shouldn't push broken commits because it means git bisect breaks if you can't test the broken code in the middle of a bisect
    • standardizing "please write to the list" notes? Possibly with a new wiki page with a longer explanation?
    • LTS Timing
      • From last time: 5.4 and 5.7 will be LTS
      • How long is Long?
      • Dave's proposal: keep 2 LTS branches active, ensuring at least 5 years. Keep 3 branches otherwise
      • Examples:
        • 5.4L, 5.5, 5.6, 5.7L
        • 5.4L, 5.6, 5.7L, 5.8
        • 5.4L, 5.7L, 5.8, 5.9
        • 5.7L, 5.8, 5.9, 5.10L
  • Technical
    • the infamous oid typedef issues
      • 5.6.1.1 still scheduled to go out
      • Should we consider going to a 6.0, where we change all types to match (the biggest becoming all the u_longs used everywhere for ints).
        • Wes notes that it may be possible to have a injectable handle that swaps types in the middle of a handler chain for legacy code
      • general discussion tabled for future meetings about 6.0
    • Windows Builds
      • Dave hopefully has windows working? Needs wiki post about doing it.
  • Round Table
    • dts12: 5.x.4
    • hardaker: 5.7 and git
    • bvanassche: wiki pages about git