HISTORY.md

Revision history

1.6.1

• Replaced npm package later back by the equivalent atmosphere package after encounting major problems on the client side

1.6.0

• Replaced meteor package later by the equivalent npm package
• Package updated to CoffeeScript 2. Now is required to create an instance of the class with new JobCollection, just calling JobCollection does not work anymore. Same for new Job.
• Replaced tinytest by meteortesting:mocha to run tests including code coverage

1.5.2

• Fixed an issue under some circumstances when job.cancel() was run on a job with option { dependents: false }. Thanks @gbhrdt.

1.5.1

• Fixed issue introduced in 1.5.0 which caused processJobs() with arrays of jobType strings to fail a new type check. Thanks @danielparas.

1.5.0

• Added errorCallback option to processJobs(). This gives workers a way to log network and other errors and do something other than write them to console.error (the default and previous behavior).
• Added workaround for bug/feature in later.js that caused the first run of periodic jobs to sometimes occur immediately after scheduling. Thanks @mitar.
• Progress information is now preserved in the job document on 'done' or 'fail', previously it was set to hard coded values. Thanks @mitar.
• Removed section recommending capped collections for managing removal of old finished jobs. This no longer works as of MongoDB 3.2. Thanks @mcoenca.
• Jobs that complete before their dependents have even been saved will no longer result in unresolved dependencies. Thanks @serkandurusoy.
• Documentation improvements
• Updated package dependencies

1.4.0

• Added support for new "callbackStrict" option to processJobs(). When true (default false) processJobs() will throw an error if a worker function calls its callback more than once. Previously it only wrote a message to stderr in all cases. That functionality is preserved with this change.
• Added job.repeatRetries as an optional (for backward compatibility) attribute of the job document model. This attribute is used (when present) for the number of retries to attempt for the new instance of a repeating job. When missing (such as will be the case for jobs created by older versions of meteor-job) the old method of using job.retries + job.retried is used.
• Fix an error caused by a check on retries when a waiting/ready job is cancelled, restarted and then refreshed. This condition also occurs when job.restart() is called with option retries greater than the current value of job.retried (which is the default case after a cancel of a non-running job.) Thanks @huttarichard.
• In coordination with the above change, the retries option of job.restart() will now accept a value of zero.
• Fixed race condition in processJobs() that could cause Job.getWork() to run after the queue is paused or stopped, leading to a zombie job on the server.
• Documentation improvements
• Updated package dependencies

1.3.3

• Fixed error affecting all of the *Jobs functions which caused them to only process the first 256 jobs in the provided list of _id values.
• Updated package dependencies

1.3.2

• Fix broken links in README

1.3.1

• Updated package dependencies
• Documentation improvements (typos)

1.3.0

• Added delayDeps option to job.done() to allow dependent jobs to be delayed after the completion of an antecedent job. Thanks to @a4xrbj1 for suggesting.
• Fixed bug in Job.jobStatuses, Thanks @niceilm
• Invoking server-only or client-only methods in the wrong environment now generates a warning method. Thanks @aardvarkk.
• Additional indexes added to make built-in queries more efficient for large collections. Thanks @mitar.
• Internal improvements in event handlers. Thanks @mitar.
• Documentation improvements, Thanks @rhyslbw, @mitar and @jdcc !
• Updated job subproject to v1.3.2

1.2.3

• Fix issue when very long promote() cycles are used, causing waiting jobs that are ready to run at a server restart to be delayed one full promotion cycle. Thanks to @KrishnaPG for reporting this.

1.2.2

• Fixed bug in the jc.getWork() workTimeout functionality that could cause a running job to immediately auto-expire on the server if it was previously run with a workTimeout value, and was then subsequently run without a workTimeout value after the original job was rerun, restarted, retried or repeated.
• Updated job subproject to v1.3.1
• Updated README to include links to job-collection playground app and node.js worker repos

1.2.1

• Fixed log() on jobs without runId by loosening check in jobLog(). Thanks @sprohaska for the PR!

1.2.0

• Added ability for workers to specify a timeout for running jobs, so that if they crash or lose connectivity the job can automatically fail and be restarted. getWork() and processJobs() now each have a new option workTimeout that sets the number of milliseconds until a job can be automatically failed on the server. If not specified, the default behavior is as before (running jobs with no active worker need to be handled by the developer.) Thanks to @aldeed for the idea of making this a worker setting.
• Added repeatId option to job.done() which when true will cause the successful return value of a repeating job to be the _id of the newly scheduled job. Thanks to @tcastelli for this idea.
• Added jc.events, which is a node.js Event Emitter allowing server code to register callbacks to log or generate statistics based upon all job-collection DDP methods. There are two main events currently implemented: 'call' for successful DDP method calls, and 'error' for any errors thrown in such calls. There are also events defined for each of the 18 DDP methods (e.g. jobDone or getWork). These events emit for every such call, whether successful for not.
• Added new DDP method methods job.ready() and jc.readyJobs() to provide a standard way to promote jobs from 'waiting' to 'ready'.
• Throw an error when a buggy monkey patch has been applied to Mongo.collection (usually by some other package). See this discussion for details.
• The built-in default server logging mechanism (through setLogStream()) has been refactored to be built on top of the new Event Emitter mechanism. It otherwise works exactly as before.
• Providing a falsy value of option pollInterval when calling Job.processJobs() will now disable polling in favor of using q.trigger exclusively.
• Fixed bug where q.trigger() caused a getWork() call, even when the queue is paused.
• Fixed a bug in job.rerun() that caused it to fail if called with a later.js object for the wait parameter.
• Documentation updates and other improvements, with thanks to @skosch for a documentation PR.

1.1.4

• Fixed issue when using multiple job-collections with different DDP connections in use. Thanks to @mart-jansink for reporting this issue.

1.1.3

• Documentation improvements. Thanks @kahmali!
• Added explicit package version to onTest() call to work around a Meteor issue when running meteor test-packages within an app.
• Fixes issue when getJob() called with getLog or getFailures options.
• Fixes incorrect log level of 'warn' for job cancellation (correct is 'warning').
• Thanks to @kingkevin for PR fixing the above two issues.
• Added a client polyfill for Function.bind() which is missing from phantomJS

1.1.2

• Fixed a bug that caused server-side calls to job collection using callbacks to throw rather than properly propagating errors via the provided callback.

1.1.1

• Updated meteor-job package to 1.1.1, fixing a bug that could cause JobQueues to get more work than configured when using q.trigger() or very short pollIntervals.

1.1.0

• Support for using later.js schedules to configure repeating jobs.
• Removed unnecessary server-side console output.

1.0.0

• jc.startJobs and jc.stopJobs have been renamed to jc.startJobServer and jc.shutdownJobServer respectively. The old versions will now generate deprecation warnings.
• jc.makeJob() and jc.createJob() have been deprecated in favor of just calling new Job(...)
• Fixed an issue similar to #51 on the client-side.
• Fixed issue #55. All standard Mongo.Collection options should now work with JobCollections as well.
• Updated versions of package dependencies
• Fixed issue #41. The potential race condition in getWork is handled internally
• Fixed issue #57. Default MongoDB indexes enabled by default
• Fixed issue #55, all valid Mongo.Collection options are now supported. However, transformed documents may fail to validate unless "scrubbed". More work needs to go into documenting this.
• Fixed #28. Eliminated all "success" console logs.
• job objects now have job.doc readable attribute
• jc.jobDocPattern can now be used to validate Job documents.
• j.refresh() is now chainable
• Added jq.trigger() method to provide a mechanism to trigger getWork using an alternative method to pollInterval
• job.log() can now accept a data option, which must be an object.
• log.data field is now permitted in the Job document model.
• When job.fail(err) is used, the error object stored in the failures array will have the runId as an added field.
• connection option to new JobCollection() on client or server will now direct the local Job Collection to connect to an alternate remote server's Job Collection rather than using the default connection (client) or hosting a collection locally (server).

0.0.18

• Fixed issue #51, which caused errors on the server-side when multiple job-collection instances were used.

0.0.17

Note! There are some breaking changes here! Specifically, the job.fail() change below. See the docs for specifics.

• Added support for Meteor 0.9.x style packages, including name change to accomodate
• job.fail() now takes an object for error. Previously this was a string message.
• Refactored JobCollection classes
• Jobs that are ready-to-run go straight from waiting to ready without waiting for a promote cycle to come around.
• Don't check for after and retries in getWork()

The following two features are experimental and may change significantly or be eliminated.

• Allow the server to add data to job._private that won't be shared with a client via getWork() and getJob(). Be sure not to publish cursors that return job._private!
• Added support for @scrub hook function to sanitize documents before validating them in getWork() and getJob()

0.0.16

• Updated to use the latest version of the meteor-job npm package

0.0.15

• Added until option for job.repeat(), job.retry(), job.restart() and job.rerun().
• Added jc.foreverDate to indicate a Date that will never come
• Fixed bug where jc.forever was not available
• Default Date for job.after() is now based on the server clock, not the clock of the machine creating the job. (thanks @daeq)
• Added created field to job document model to keep track of when a job was first created on the server

0.0.14

• Added idGeneration and noCollectionSuffix options to JobCollection constructor. Thanks to @mitar for suggestions.
• Removed unnecessary console log outputs
• Updated README to point to new sample app, and to clarify the use of jc.promote() in Meteor multi-instance deployments.

0.0.13

• Fixed bugs in client simulations of stopJobs and startJobs DDP methods

0.0.12

• Fixed bug due to removal of validNumGTEOne

0.0.11

• Fixed bug in jobProgress due to missing validNumGTZero

0.0.10

• Changed the default value of job.save() cancelRepeats option to be false.
• Fixed a case where the echo options to job.log() could be sent to the server, resulting in failure of the operation.
• Documentation improvements courtesy of @dandv.

0.0.9

• Added backoff option to job.retry(). Implements resolves enhancement request #2

0.0.8

• Fixed bug introduced by "integer enforcement" change in v0.0.7. Integers may now be up to 53-bits (the Javascript maxInt). Fixes #3
• Fixed sort inversion of priority levels in getWork(). Fixes #4
• Thanks to @chhib for reporting the above two issues.

0.0.7

• Bumped meteor-job version to 0.0.9, fixing several bugs in Meteor.server and Meteor.client workers handling.
• Corrected validation of jobDocuments for non-negative integer attributes (integer enforcement was missing).
• jc.promote() formerly had a minimum valid polling rate of 1000ms, now any value > 0ms is valid
• Added a few more acceptance tests including client and server scheduling and running of a job.
• Documentation improvements

0.0.6

• Added initial testing harness
• Fixed issue with collection root name in DDP method naming.
• Changed evaluation of allow/deny rules so deny rules run first, just like in Meteor.
• Documentation improvements.

0.0.5

• Really fixed issue #1, thanks again to @chhib for reporting this.

0.0.4

• Test release debugging git submodule issues around issue #1

0.0.3

• Fixed issue #1, thanks to @chhib for reporting this.

0.0.2

• Documentation improvements
• Removed meteor-job subproject and added npm dependency on it instead

0.0.1

• Initial revision.