You are not logged in Log in Join
You are here: Home » Members » poster » Zope Asynchronous Call Manager » zasync-1_1 » Change file » View Document

Log in
Name

Password

 

Change file

Changes to zasync through 1.1

version 1.1.0 (tag zasync-1_1_0)

(note that there is a spurious zasync_1_1_0 tag as well)

  • Add Windows files and rc files from Arno Gross into bin directory.
  • Put more information on diagnostic call analysis page, thanks to Arno Gross.
  • Make zasync work in 2.7 and 2.8. Note that making it run in 2.7 now requires an extra argument to run: see the example files in the bin directory. Briefly, here are two example calls, from one of my test installations:

    For 2.7:

    ~/zope2/bin/python -c "import zasync; zasync.run(/home/gary/zope2/instance277/etc/zasync.conf, 2.7)"

    For 2.8:

    ~/zope2/bin/python -c "import zasync; zasync.run(/home/gary/zope2/instance/etc/zasync.conf)"

  • Add some example text about the zope_exec plugin in plugins.txt.

version 1.0.7 (tag zasync-1_0_7)

  • Some small changes (like using get_transaction().begin()) driven more by superstition and risk reduction than anything else.
  • Make manager include lines that make Python datetime objects acceptable to use in restricted python. This should make it easier for a simple install.
  • Fix edge-case bug: if you have a call manager with new calls that have expired, the client would not be working with wrapped deferred objects so aq_... attributes would be missing and the new call wouldn't be tossed, and new (non-expired) calls wouldn't be processed.

version 1.0.6 (tag zasync-1_0_6)

  • More fixes for the code path of handling error-raising reprs (grr).

version 1.0.5 (tag zasync-1_0_5)

  • cleanFailure using customized code because __repr__ can fail in Zope. :-/
  • Fix nasty small bug in client.py that could make zasync effectively into a zombie process. :-(

version 1.0.4 (tag zasync-1_0_4)

  • Add in ability to pass in form variables to request.
  • Fix bug in rendering error message in client config.

version 1.0.3 (tag zasync-1_0_3)

  • fix retry code in client to call makeCall correctly

version 1.0.2 (tag zasync-1_0_2)

  • Try for more and earlier transaction aborts.
  • Reinstate lost diagnostic call analysis page.

version 1.0.1 (tag zasync-1_0_1)

  • Added new diagnostic call analysis page for the ZMI call manager. Unfortunately it was not exposed in this release. :-/
  • Attempt to improve the README file to give people more of a chance to get this set up and going.
  • A clean up of the two example configuration scripts to remove references to Zope 3, which really probably just confused matters.
  • Manager needs to fix up local roles to get the owner info in.

version 1.0 (tag zasync-1_0)

  • Add an overview tab that includes information on the plugins and allows you to ping the zasync client to see if it is active.
  • Make the verbose traceback optional, defaulting to False. Incorporate the new verbose traceback option in logging output
  • Make the normal Zope event log work.
  • Go ahead and include temporary storage in schema and example zasync.conf so DBTab doesn't complain.
  • Add new "aggregate" plugin.
  • Include exception info in Conflict Error logs.
  • Make a returned failure in a zope exec callback abort the transaction.
  • Bug fixes in plugins, especially zope_exec
  • fix bug: if tool disappeared, then zasync was unable to recover because of a number of problems. Most importantly, it was holding on to connections without closing them. Fixed this bug and similar problems elsewhere.
  • More paranoid about starting up now, and more paranoid about telling the manager about the available plugins when the tool has been re-found.
  • tweak backoff interval for assigning tasks; add backoff interval to calling zope_exec start_worker so that existing threads have a better chance to grab new jobs before we create an unnecessary new thread.
  • Add sanitize method so that calls to (and from zope_exec) do not have persistent objects.
  • Remove an opportunity for Conflict Error in the client.
  • Don't raise an error when a deferred is called twice--this will abort the transaction, which means that the deferred will never be resolved if it was supposed to be. Therefore, return a failure and log the problem.
  • Add a here to the context dict for the callbacks because some code wants it.

version 0.2

  • Significantly rewritten to solve the problems of 0.1.
    • don't keep persistent objects around
    • try to handle exceptions better
    • close the connection after every transaction
  • remove all CMF dependencies of the call manager
  • Use a ZConfig-based configuration for zasync
    • better Zope initialization
    • configuration-driven plugin registration
  • saner startup
    • not reliant on zopectl run (intended to be in preparation for using twistd)
    • sh start script
  • try to remove plugins from the manager's plugin list upon zasync shutdown
  • ZPL

version 0.1

Initial version. Not released. ZVSL.