This is due to the byte positions being taken prior to the input being formatted for display

In streams-> data tab
press upload button
when dialog appears click in the effective date box
when calendar appears press enter
get NPE

(in v4 the above steps cause a blank stream to be uploaded)

stroom.security.server.AuthenticationServiceMailSender.()

throws java.lang.NoSuchMethodException

Currently deleting a folder that contains entities fails with a FK violation. Need to give the user the option to delete all the contents as well (including any sub folders) or cancel.

If the selected entity is changed since the DocRef was created then the name shown will be incorrect.
If the DocRef contains no name for the selected entity then it will appear as if no entity has been selected (blank text box) even though the selection is valid. This is especially confusing post upgrade where all of the entity references may appear as if they have been wiped out.

This is related to when a proxy combining several data sets in to a "batch" of data.
When a batch contains multiple smaller data sets for the same FEED, only the first dataset can be "debugged" in the play forward function where you select the RAW_FEED and then the pipeline you want to debug/test/develop.

Table should treat blank columns as empty strings

An expression could be used to create the link text and the link could open another dashboard with parameters or an external resource.

Add a window closing handler to Stroom that checks to see if any content tabs have dirty (unsaved) content and asks the user if they really want to leave/refresh the page if there is any unsaved content. It may be good to ask the user if they really want to leave even if there is no unsaved content as they could be in the middle of a query or other activity.

An example of a window closing handler used for dashboards that could be added to ContentManager:

 Window.addWindowClosingHandler(event -> {
           if (dashboardPresenter.isDirty()) {
               String name = "";
               if (dashboard != null) {
                   name = "'" + dashboard.getName() + "'";
               }

               event.setMessage("Dashboard " + name + " has unsaved changes. Are you sure you want to close it?");
           }
        });

If the above is added then the KeyboardInterceptor can be changed to stop blocking F5 and backspace keys as this is crude and potentially problematic.

Create a process on stroom startup to import all content packs found in a configured directory.

Intended for use on new installations or docker instances so that stroom can startup with the required content, e.g. core schemas, template pipelines, etc.

Create a property to enable/disable this feature.

Mark the property as disabled after first run.

Move the imported packs into a sub directory and failed packs into a failed sub-dir.

When a new instance is started up the user must create a volume before they can store any data. For docker instances and the startup of a new instance it would be useful if stroom created a default volume on the root filesystem.

This could be disabled via a property for production installations.

If you set the XSLT to ${feed}

I have several different Raw Events [Audit Raw Events] [List Raw Events] etc.
Currently all Raw Events and translated Events are stored under the same path.
I would like to store (any) Raw Events on a slow file system and all Event data on a Fast file system such as SSD.

I.e.
/stroomdata/Raw_Events/Audit_Raw_Events/hostname/etc
/stroomdata/Raw_Events/List_Raw_Events/hostname/etc
/stroomdata/Events/hostname
Where /stroomdata/Raw_Events is mounted to a slow file system and
/stroomdata/Events is mounted to a fast file system

Cheers

In the visualisations, currently sorting of series is done by the name of the series. We need the ability to sort by one of the values within that series, e.g. if one of the fields is a count then doing a sort DESC by that count would give us a top-n type query.

The setting dialog would need a drop down (or some other mechanism) to allow the user to specify the field that the sort will operate on.

When using a date field in tables and queries containing zone data, neither queries, formating of the field or formulas on the field works. If the zone data is UTC, i.e. 0, it works fine.

The order of the fields on a query does not match the order of the fields in the index.

Allow dashboards to be created with conditions in the query for which the user will be prompted when the dashboard is run.

This would make it easier to do things like get a quick view of events from a system. E.g. run and just pop in the name of the system and optionally event type then see results. Would feel less cludgey than changing an existing dashboard.

Ideally...
Values that are not filled in by users would have their conditions removed before running (i.e. don't error)
Default values could be assigned during dashboard creation
When the query is re-run, the previously used values are utilised, maybe with a "don't show this dialog again" to always use the same values for subsequent runs.
The name of the dashboard on the tab would be able to be composed from the supplied parameters, e.g. to make it easy to refer back to tabs if several dashboards have been run up using the same template.

Even if you include a "\n" in the string it does not create a new line.

If I use the function under expression to trigger a true or false value, I'm unable to filter in include/exclude using the true or false value.

At the moment, it is possible to filter on several different fields such as streamid, date etc. There is an area where it is possible to filter on some of the meta data.
Please add the ability to filter on any meta data supplied along with a batch of data to a Feed. I.e. it would be really handy to be able to filter on sending host, the UUID of the dataset etc.

Hessian is generating a StackOverflowError when serialising the search object when it is trying to contact worker nodes.