Spec

We want to add support for following repeat group button labels on mobile -

 <itext>
        <translation lang="en" default="">
          <text id="$repeat_group_id-add-new-label">
            <value>Repeat</value>
          </text>
         <text id="$repeat_group_id-add-another-label">
            <value>Repeat</value>
          </text>            
        </translation>
</itext>

The current naive strategy for downsampling captured images doesn't perform well when resizing images with light patterns, as they end up very stippled. We should experiment with different platform methods for resizing which don't amplify these underlying shadow overlays

There is a bug in automated image resizing when an image's existing / native size size is exactly the width of the screen.

Discovered with a specific image and replicated with the following virtual device profile:

Adding this 456x456 image as the multimedia image for a question label with the above device screen parameters should result in a crash. I can't quite remember if the bug was when this image was 456x456 or 480x480 (exact screen size or size minus padding maybe), but resizing to one of those should work.

Further Context: https://dimagi.atlassian.net/browse/SUPPORT-16630

Entity lists currently support fuzzy matching as a global option, but fuzzy matches are a poor fit for some types of search results like QR code scans.

To support this, individual fields should probably support their own tags for what types of matches are relevant / desirable. Fields with canonical inputs rather than 'human' inputs should have different matching criteria.

Relevant user feedback discussion

Hello!

This is Milen from Simprints. I'm currently working on an integration with CommCare's Case Content Provider and I believe I have found a bug in it's implementation. Following this documentation I'm trying to access the case metadata listing and make a selection based on the owner_id. This is my code:

selection = "owner_id = ?"
selectionArgs = arrayOf(decryptedAttendantId.value)
val caseMetadataCursor = context.contentResolver.query(CASE_METADATA_URI, null, selection, selectionArgs, null)

However, I'm getting a null cursor and looking at the logs from CommCare I see the following exception:

java.lang.RuntimeException: Invalid selection key for case metadata: owner_id

Digging in your code I found out that the OWNER_ID column is actually defined as "owner_ID". I tried passing "owner_ID" in my selection but got the exact same exception. Digging further in your code I found this line which converts all selection keys to lowercase. This essentially makes it impossible to make a selection on the "owner_ID" column.

Let me know if I'm missing something or if you need any more info!

Turning Strict mode with death penalty currently causes app to crash at several places. There does not appear to be much logging around why these issues are occuring. We should look more into these violations and try to resolve them if possible.

Clean up following J2ME properties from test files (these should not be getting used anywhere in live code) -

• properties.password_format  # Password Format
  

• properties.cc-send-procedure  # Send Data
  

• properties.server-tether  # Sync Mode
  

• features.sense  # CommCare Sense
  

• properties.extra_key_action  # Extra Key Action
  

• properties.cc-entry-mode  # Item Selection Mode
  

• properties.cc-send-unsent  # Send Forms Mode
  

• properties.ViewStyle  # Form Entry Style
  

• properties.cc-user-mode  # User Login Mode
  

• properties.cc-login-images # Login Buttons
  

Please update Bouncy Castle

• https://www.bouncycastle.org/
• https://www.bouncycastle.org/releasenotes.html
• http://www.bouncycastle.org/latest_releases.html
• https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=bouncy%20castle
• https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=bouncycastle
• https://www.cvedetails.com/vulnerability-list/vendor_id-7637/Bouncycastle.html

• Fix Allow GPS popup on opening a form.
• Browserstack sometimes automatically re-open wifi, after we manually disableWifi causing tests to fail.
• Update FormSubmissionTest to directly use the submitted form instance ID for receiving acknowledgement.
• Fix and revert #2381

For more security, can you change old spongycastle (based on old bouncycastle) to bouncycastle?

• https://www.bouncycastle.org/
• https://www.bouncycastle.org/releasenotes.html
• http://www.bouncycastle.org/latest_releases.html
• https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=bouncy%20castle
• https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=bouncycastle
• https://www.cvedetails.com/vulnerability-list/vendor_id-7637/Bouncycastle.html

The Case List Map feature supports a single roaming indicator of the user's current position, based on the current location reading from the device.

This single geopoint is helpful but limited in helping users to most effectively use the map for 'homing' tasks where the user is attempting to use the map to physically navigate to a precise GPS point.

The following would be strong additions:

• Include an accuracy circle which can communicate the uncertainty in the user's current location
• When accuracy is sufficiently poor (above some threshold), remove the 'dot' indicating the user's current location, to prevent them from misinterpreting it as an accurate depiction
• Provide a pop-over banner on the screen when a user's location is sufficiently inaccurate, to inform them as to why they can't see their precise location
• Support for the "compass" functionality otherwise available in google maps, in which a cone is thrown based on the direction the device is facing, allowing users to disambiguate between equally close points

Currently we use ModernHttpRequesterMock in unit tests to test out the http requests. We should replace this with OkHttp MockWebServer in order to make tests more dependable by relying on an universal mock library rather than something that we setup in our tests. This will allow us to test network requests directly with okHTTP layer which is the ultimate source of truth for us for network requests.

We would like to deprecate cc-enforce-secure-endpoint property, though we are not sure if there are self hosters utilising this. We should log any usage of this property in Analytics to be able to verify if it's getting used by anyone.

When image resizing is used in captured images, EXIF Data is currently stripped. Since most Cameras in Android don't natively rotate the pixels in a photo, but rather set the orientation captured in EXIF, this means that resized images don't display in the same orientation as the original captured photo.

We should either pass-through some EXIF info during resizing, or read the orientation flag and manually rotate the captured image.

Right now we solely use the embedded zxing library, but there are circumstances where users might want to jump out to a much more powerful version that's more appropriate for their device or use case.

The need for this would be negated by implementing #2321, but that's potentially a significantly larger scope of work compared to a fallback option for telling the barcode scanner to use the traditional zxing intent, rather than the built-in one.

This seems to me like an App setting rather than a callout specific setting, so I think the best course of action would be to make it an application preference ("Use intent callout for scanner") that is loaded in the FormEntryActivity and passed into the widget factory .

The image capture process creates a new copy of the source image without passing through any existing captured EXIF properties. In most cases this makes sense because details about the captured image (Camera, resolution, etc) are no longer meaningfully true, but most notably this means that GPS / Geolocation data is stripped.

There isn't total consistency in what EXIF properties exist and are valid, but the common tags for location data could be retained without significantly concern. It would be worth a quick look to see if there are any other common properties that users may want to retain.

• DeletedUser
• FormEntry
• Fixtures
• MenuTests
• Login
• FormSettings
• CaseSharing
• Languages
• CaseListSearch
• CaseListSort
• Settings
• FormFiltering
• Tiles
• SessionExpiration
• Dialogs
• InstallList
• DemoUserOnline
• DemoUserOffline
• LookupTableOptionSorting
• ManualQuarantine
• LogSubmission

Currently the String Widget doesn't support EditorInfo.IME_ACTION_NEXT like numeric entry widgets, since it's mixed use for both "short" text entry (IE: Names on one line) and for long (multiline paragraph entry) cases.

This is very unfortunate for the 99% use case of short text entry. It would be great to support both, and eventually transition the default behavior to be 'short' once people have had enough time to adjust defaults.

Hello,
As I import commcare android in android studio I am not able to build project because of the following issue
Unable to resolve dependency for ':app@standaloneRelease/compileClasspath': Could not resolve io.ona.kujaku:library:0.7.2.
Could not GET 'http://dl.bintray.com/ona/kujaku/io/ona/kujaku/library/0.7.2/library-0.7.2.pom'. Received status code 403 from server: Forbidden
Please suggest
Thanks
Siddesh S

The IntentCallout activity assumes that called services have knowledge of ODK's incredibly specific expectations. Namely that the result be bundled and named odk_intent_data. This makes the callout feature unusable when trying to receive a result from most applications.

Would it be possible to see if odk_intent_data is actually returned, and if not try to getString() for the available keys that match the requested keys? Even if you can only handle strings, it would go a long way to making generally IntentCallout useful.