• Be able to remove or not the log output for test results (should improve testing loops)
• Keep max and min times on the internal __profile
• toMarkdown should aggregated results for multiple runs

Be able to bulk check the versions of all installed opacks and update them.

A useful helper would be an argument type and data validator.

When using oPack genpack for yaml package file generation the logging seems to leave unwanted characters once the command finishes.

This opack command for script, daemon and ojob documentation should be documented and tested after a first phase of tests.

When using a DB object with convertDates(true) and the SQL query returns a null date an exception is thrown. A possible workaround is to use a NVL like function on SQL.

Support for basic authentication when dealing with remote opack operations (e.g. installations, upgrade, check, etc...)

Optionally it should be possible to audit all authorized (and none anonymous) access to an OpenAF channel. This should include the info for which key was changed/deleted by a specific access.

Rewrite repacking procedure in Java for better compatibility.

In oJob, whenever the same job appears twice on a todo list, it should be handle either be creating an extra execution for that job or ignoring and warning that a new job should be added to provide the same functionality:

jobs:
   - name: myjob

   - name: job1
     to: myjob

   - name: job2
     to: myjob

jobs:
   - job1
   - job2

The -e option or a new option should be added to allow for direct script commands.

When using functions like DB.qs parameters the possible to use Dates.

The nightly build produces the following error when starting the openaf-console:

java.util.logging.LogManager$RootLogger log
SEVERE: Failed to resolve default logging config file: config/java.util.logging.properties

In the following situation:

var a = Packages.org.apache.commons.io.IOUtils.toByteArray(anInputStream);
var s = af.fromBytes2String(io.gunzip(a));

The io.gunzip will receive a as a NativeJavaArray instead of a B[]. So the NativeJavaArray needs to be "unwrapped" in both io.gunzip and io.gzip if instance of NativeJavaArray.

Add more functions to provide easy access to memory profile on ow.test and add them to odoc.

Showdown is already included as a javascript library. Now it just needs to add one or two markdown css bundle on the css folder and corresponding owrap.server and/or owrap.template to render markdown to html.

The load and loadLib functions should be reviewed further on how they make a distinction between FileNotFound exceptions. In some cases it's difficult to assess which is the cause of loading failure.

A job could be triggered by channel subscription like in nattrmon (OpenAF/nAttrMon@8ba5bf5).

The automatic tests for ow.ai need improvement as sometimes they produced false "fail" results due to the test case and conditions verified.

When you have a set of opacks on the same folder as openaf.jar, upon install it will try to install all of them. But dependencies are not consider or not searched on the same path.

When executing a command with sh/af.sh that outputs a "large content" (>1KB) the sh/af.sh seems to get stuck (even with timeout). But if executed with inheritIO = true it works perfectly.

Solving or working around Java 9 GA compiling and execution issues:

• Automated tests fail for gzip native to byte array
• Automated tests warnings on read/writeFileStreamBytes
• Deprecation warnings on building
• Repack operation fails
• jdeps -jdkinternals openaf.jar
• Rhino issue: mozilla/rhino#352

When using the DB object and setting convertDates(true) the timestamps fields are not being properly converted. Instead the value is shown as "oracle.sql.TIMESTAMPTZ@1234567". As a possible workaround a trunc is enough to convert back to a date.

Implement a channel type using H2 MVStore with options to be persisted in memory or disk, map with a single name or based on a function, optional map housekeeping with a function, compression options, etc.
Then probably some ow.ch.utils functions to provide map names and housekeeping based on date and housekeeping based on number of records. This would be useful for a new output on nattrmon.

Study the possibility of using Apache HttpComponents for some core functionality instead of using Java's URLConnection.

Need to move the old autoTestAll.js tests on ow.* functionality to the new auto test ojob:

• ow.ch
• ow.format
• ow.obj
• ow.oJob
• ow.server
• ow.template
• ow.test

The openaf-console SQL table color seems to have changed to black instead of the previous color.

When using ojob -jobs example.yaml if there are ojob includes the ojob list will include duplicated jobs.

The internal log housekeep in oJob tries to clean entries older than 1000 but it's not restricting only to entries that have successfully finished or error. This should be changed despite being a very rare situation.

When connecting to a new host that isn't on the .ssh/known_hosts you might not be able to connect in some cases. This can usually be workaround by connecting one time with the command-line ssh command. But it would be helpful if the plugin could add a host to the known_hosts (using something like https://dentrassi.de/2015/07/13/programmatically-adding-a-host-key-with-jsch/).

Whenever .opack.db loses reading permissions for the user running a openaf script using libraries from opacks, there isn't any warning or exception to remember that .opack.db has the wrong reading permissions.

The UNIX shell script and windows batch that are generated during --install should test for java and use the path available java if the one detected on install is no longer available.

Study the possibility of moving from JSch to SSH-2 without breaking existing functionality in the SSH plugin.

While loading an .openaf-console_profile if there are errors executing it they aren't displayed as in .openaf_profile.

When creating or updating an opack it should use saveHelp and check an odoc entry on the .package to automatically generate the odoc database.

The default behaviour is for a job that has dependencies to only execute the exec block if and when all dependencies are successfully executed. But when a job dependency fails the job will wait for any successful re execution of the job dependency. It would help if a exec block for onFail of a dependency could be created and also decided (by returning false or true?) if the job should wait for a re execution or also fail.

 - name      : My results
   deps      :
       - name     : My first job
         onSuccess: print("Great, first job done.");
         onFail   : print("I really need the first job"); return true; /* return true for the "My results" job to fail */

       - name     : My wrong job
         onSuccess: print("Great, wrong job went ok.");
         onFail   : print("Wrong job failed but actually it's optional. Just to warn you."); /* returns false by default so "My results" job doesn't fail */

    exec      : |
       if (args.__deps['My wrong job']) { ... }

When using ow.ch.utils.setLogToFile it will gzip log files when using JVM 1.8 and won't do it in JVM 1.7.

The sync in ow.template handlebars needs to be removed for various reasons (e.g. performance, ugly, etc...)

It would be helpful to have autocomplete of alias in the openaf-console.

When using:

ojob file.yaml -deps

the dependencies area is not interpreting correctly the onFail/onSuccess job dependencies entries.

Add a generic ojob option to log to files instead of console using ow.ch.utils.setLogToFile.

When using ow.ch.getLogStashSubscriber in some situations it tries to log a string date instead of a JSON date.

Study the possibility of moving from streamjs to sequency without breaking $stream.

After the last revision of ow.template to use require the addFormatHelpers and addConditionalHelpers no longer work as expected since a subsequent call to parse, for example, won't consider these helpers. Additionally more automated tests for this should be added.

As a workaround you can use HandleBars directly but it kinda defeats the propose.

After creating a remote channel there should be a proper "close" function. This function should do all the necessary cleanup including sending HTTP 401 if authenticated.

Executing "opack install" should return exit code 0 instead of a non-zero code. This would be specially helpful when building docker images, for example.

As the js folder increases in size it would be better organized if it was divided into sub-folders. That means changes on the build and load process to be assessed.

The $ch storeAdd, storeRestore and createRemote create the corresponding channel if it doesn't exist. expose and peer should have a similar behaviour. This was expected on ojob expose channels listing, for example.

Currently ow.ch.utils.getElasticIndex produces yyyyMMdd but it should default to yyyy.MM.dd plus allow to use a template.