Megatest: Update of "Releases"

Overview

Content

Version 1.44 May 4, 2012

Better filters in dashboard, e.g. "run1,run2.%" will show "run1" and all runs matching the pattern "run2.%"
MT_TARGET variable available in runs
Added "toplevel" tests. Create a test with a waiton that specifies the tests you wish to run and put "mode toplevel" in your [requirements] section. Running this test will run all dependent tests and finally the toplevel test itself (can be used to summarize the run).
Fixed the run blocking bug where failed tests caused the running process to block and never exit.
Unit tests have been re-enabled using the -repl command.
Detect the case where a waiton specifies the calling test (would previously go into an infinite loop).
The beginnings of a user manual written in asciidoc instead of Lyx. See docs/manual.
Added a simple generator for a megatest regression area and for new tests. Try "megatest -gen-megatest-run" and "megatest -gen-megatest-test"

Version 1.44 May 4, 2012

Better filters in dashboard, e.g. "run1,run2.%" will show "run1" and all runs matching the pattern "run2.%" * Version 1.41 Apr 3, 2012

Link path stored in db instead of runpath
Debug settings passed to running tests. WARNING: using this can cause false fails if you process STDOUT as part of your test
Run areas and links to them have been cleaned up.
Many speed-ups and db read/write optimizations
Addition of #{ .. } mechanism to config files to accommodate advanced needs.

Version 1.39 Mar 26, 2012

Caching of writes to db in server mode
Streamlined many writes to not do redundant lookup of test-id
From dashboard remove runs process is kicked off in an xterm
In ezsteps scripts no need to specify ./ on systems with . not in path
Trim tests from to-run list to speed up starting new tests (not constantly checking status on bunch of completed tests)
Addition of nbfind (can be used instead of nbfake) which will run locally if load is below 70% where load is defined as L/Ncpu where L is unix load and Ncpu is the number of cpus.

Version 1.23 Sept 6, 2011

NB// this release changes the db schema. You should make a back up copy of your megatest.db and run “megatest -rebuild-db”

Run rollup. A group of partial runs can be rolled up into a single run. All records are copied to the newly created run and run deletion has been updated to not delete run directories where there are still references in the database.
Arbitrary data roll up from tests. From any test data can be rolled up into the database that will then be available when the database is dumped to csv. There is no limit imposed on how much data can be rolled up but the current database used by megatest, sqlite3, will slow down with very large data sets (over 10 meg or so per test).
A scroll bar mechanism has been added to the dashboard.

Version 1.22

Waiver propagation. Waivers will carry forward in a series of runs where the keys are the same.
Rollup of test result value, expected value and tolerance.

Version 1.04 (May 10, 2011)

Version 1.03 (May 8, 2011)

Version 1.02

Add path to megatest used at launch to the PATH variable when tests are run.
max_concurrent_runs to limit how many tests are allowed to run in parallel.
State and status for tests that fail to call home are set to INCOMPLETE and DEAD/STUCK respectively
Added -runstep for making running a single step much simpler for the common case. Also supports the -logpro parameter to use a logpro file to process the log.

Version 1.01