GitLab

Change-Id: I5ced13169a0c0a92121783f5d164a7da8e355e91

YoungAndroidProjectServce -> YoungAndroidProjectService
overriden -> overridden
invididual -> individual
Signed-off-by: Hollow Man <hollowman@hollowman.ml>

Fix typos:

propertyDefintions -> propertyDefinitions
Exsting -> Existing
Signed-off-by: Hollow Man <hollowman@hollowman.ml>

When an error occurs while saving a file we need to put the associated
file editor back in the dirty editors list. Otherwise data may be
lost. Prior to this fix once i/o failed, the editor was considered clean
and i/o would not be retried unless the editor was modified again. So
for example if a designer attribute was changed, but the i/o failed, it
would not get saved again, even if a change was made in the blocks
editor and even if the “Save project” menu was used.

Note: We do not re-schedule the failed i/o but instead depend on another
user change (anywhere) or explicit use of the “Save project” menu. I
considered re-scheduling the i/o but am concerned that if i/o is failing
due to a server problem, a lot of clients retrying i/o may make it worse
via the “thundering herd!”

Change-Id: Ife50c3c20d407b4e009c639bc36c2596331342d7

Change-Id: I6ab0a4f0e50c8c4b0ba5de47625c28723bf198cb

Change-Id: Ib6228b90ebf99c56ef8500968d61bdd60ab78324

* Implement stats collection for builds

Change-Id: If2ef82e55747522633fe2a3c458445f99332d7ac

* Load built-in templates via the repo parameter

Change-Id: I62265b3a60ecf7b651d9a22a32775e5d7434d5e5

We run 9-15 buildservers simultaneously and have a status page that
reports the /buildserver/vars foreach server. However it is hard to know
which status is for which buildserver. So we add the buildserver
hostname here.

Change-Id: I46fc59626cccf668ab57a5f81581273f0358fd1e

Change-Id: I5d7de7966f27b5ae64ea38662daa00ea12ab519e

Change-Id: I3f186175afcb254bc9a329c2a0f445fc0e42b48a

This commit is a further address to the issue #1518

Currently when the selected language is not English in login page, when login failed (e.g the password is incorrect), the login page will return with an English Page.
The commit fix this problem by adding an additional locale parameter to fail method in LoginServlet.java

Chrome 89 implements RFC8285 by default. It implements a general
mechanism for RTP header extensions. We do not use these....

Unfortunately the WebRTC library we are using in the Companion does not
understand the extension request in the WebRTC offer message and refuses
the offer completely, resulting in WebRTC not working from Chrome
89 (and presumably onward).

This change removes the extension from the offer prior to sending it to
the Companion, so WebRTC works again.

We believe this is harmless, because we do not use audio or video over
WebRTC and this extension has no impact on the data channel (which we do
use).

Change-Id: I68f4ebaa2099ca0b4bbc7b1abd0a565dbde1f4e6

The POP_FIRST_SCRIPT updated so that when the list is empty, it is set
to the empty list instead of the empty object (which appears to be the
Redis default when you empty a “table”).

Change-Id: I15002a883eead373b3d23e85867c0b1ab103c0fd

Change-Id: I00848e83a355e209f7c1e13335c21a1a42687522

Change-Id: I5338ee15a316828495e3782491a90785b46cdb1b

Change-Id: I931be1eb40f800d168aeb3b110f9a872b4e0385f

Change-Id: I11033332060ba9f9dee29fc41dd661a1e48f7081

The UpdateDone event means that a store operation (StoreValue,
AppendValueToList or ClearTags) has completed. Because some
applications, such as SketchAndGuess can cause numerous store operations
in a short period of time, there is a risk that these events may
overload the sending device. In the case of a large number of update
events, only one UpdateDone event will be triggered per tag updated
once the backlog of update operations is cleared.

The most likely use case for UpdateDone is to handle storage of large
objects. All store operations are performed in the background, so this
event permits apps to know for sure when an update operation is
complete.

Change-Id: Ia6ff5760ad622efecc9a8d235a177b9c3e00c6a9

Internally we use the Java “Jedis” library to communicate with
CloudDB. However this library is tuned for high speed low latency
connections. In particular it provides for a 2 second timeout on
transactions. This is not always sufficient in a wide area network
situations that a lot of devices will be.

Most transactions work, but when storing larger objects this 2 second
timeout can be exceeded. This change updates the timeout to be 20
seconds instead of 2. It also works around a minor issue in the Jedis
library that was causing the authentication (login) transaction to be
duplicated. This was not a real problem, but it still should be fixed,
and is in this change.

Change-Id: I00056d5bcf6f79fc783da2b22ed68c17e946e0e5

Add two boolean fields. The “monall” field (in production for a while)
permits the monitoring of all key changes. This is used for statistics
gathering and research.

The “privileged” flag permits commands that manage the overall server. At
the moment the only such command is one to manually close specified
client connections.

Note: MIT App Inventor itself does not make use of these flags, but the
MIT App Inventor source tree is the canonical authoritative location for
the token format definition.

Change-Id: If7cd4ae7cf2fd49718607aeac51d8a56f76df877

Change-Id: I305dca87f36e6d6659a3586cb9b3b014bbcafa89

* Correct order of blocks affecting global variable references
* Fix renaming of global variables when keyword translated