Change splash db upgrade logic #10731

phate89 · 2016-10-19T15:23:37Z

Change the splash screen to show the message of upgrading database only when we're actually upgrading a database.

Description

Motivation and Context

The old method wassn't aware if there was some upgrade and what db was doing the upgrade

How Has This Been Tested?

Tested upgrading my old mysql databases

Screenshots (if appropriate):

Types of change

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

My code follows the Code guidelines of this project
My change requires a change to the documentation, either Doxygen or wiki
I have updated the documentation accordingly
I have read the CONTRIBUTING document
I have added tests to cover my change
All new and existing tests passed

phate89 · 2016-10-19T15:29:14Z

@Paxxi @DaveTBlake for review/test

DaveTBlake · 2016-10-19T16:00:58Z

A welcome follow on from #10483 that only shows when migration tasks happening rather than every start up. @MilhouseVH, @tamland and @fritsch may be interested in this too.

addons/resource.language.en_gb/resources/strings.po

+#. Localized addon database name for splash screen
+#: xbmc/addons/AddonDatabase.h
+msgctxt "#24152"
+msgid "Addon"


MilhouseVH · 2016-10-19T17:22:47Z

Thanks @phate89 - it works well, I've tested upgrading MyMusic50 to MyMusic60 and MyVideos90 to MyVideos107, the latter is particularly time consuming. I'll include this PR in future builds.

There does however remain one issue: when <splash>false</splash> is added to advancedsettings.xml, there is no splash displayed, and also no migration messages (even when Music and Videos databases are being migrated). All the user sees is a totally blank (black) screen, with no messages.

IMHO the splash should be forced to appear along with the migration messages, regardless of the <splash> setting, whenever a migration is taking place.

One other nitpick would be to position the text a little higher, just in case of any overscanning or weird resolution issues - it seems a little unnecessary to position the text quite so low down (it appears to be close to the bottom edge of my 1920x1080 1:1 pixel screen).

DaveTBlake · 2016-10-19T17:43:39Z

+1
Tested with a large MyMusic48 to MyMusic60.

I too would put the text a little higher, and probably shouldn't capatalise "Database", e.g.
"Music database migration in progress - please wait..."
But minor stuff, and I would happily live with it.

phate89 · 2016-10-20T10:23:30Z

I updated the pr.. Now we update the text adding dot every second..
If someone is familiar with our CThread should check my code...

tamland · 2016-10-20T10:43:56Z

Sorry but not liking how this is currently implemented. The database should not have a dependency on the global application messenger. Instead it should be passed in as a callback/another agnostic structure, or the entire migration refactored and uncoupled from the database. This is why I didn't bothered with it in the first place.

xbmc/utils/Splash.cpp

+        Sleep(1000);
+    }
+  }
+  Show();


Paxxi · 2016-10-20T12:00:24Z

@tamland trust that passing in a callback could be cleaner but more work. This is a step in the right direction and app messenger is purposely designed for this with very few dependencies itself.
Any work that removes some lower level dependency of a higher level is progress and I encourage it. Will review the implementation later when I'm at a computer

phate89 · 2016-10-20T12:01:34Z

@tamland I added a new commit that moves the update part (the same update procedure, no refactor for now) to databasemanager. So only databasemanager has a dependency on applicationmessenger..Is it better?

FernetMenta · 2016-10-21T19:27:24Z

you could offload the migrations tasks to an extra thread. then the main thread can wait in init for this task to be completed and display splash and messages accordingly.

phate89 · 2016-10-21T23:23:01Z

@FernetMenta thanks for the suggestions. Any help is appreciated.
The main problem wasn't to actually update from main thread but update the text every second to keep something moving.
You are suggesting move the loop in capplication? Something like

CDatabaseManager::GetInstance().StartInitialize();
while (CDatabaseManager::GetInstance().IsRunning())
{
  CSplash::GetInstance().Show(updatedMsg);
  if (CDatabaseManager::GetInstance().IsRunning())
    sleep (1000);
}

FernetMenta · 2016-10-22T06:56:37Z

@phate89 yes, this will hold processing of the main thread until upgrade has finished. This is intended behaviour.

phate89 · 2016-10-22T09:34:41Z

@FernetMenta I updated the pr.. I moved the update logic to the db manager and moved the db upgrade and the addon migration to a separate thread (If I understood well our cthread)

xbmc/Application.cpp

+      if (CDatabaseManager::GetInstance().IsRunning())
+        Sleep(1000);
+    }
+  }


tamland · 2016-10-22T10:08:40Z

Not exactly thread safe.. Try creating a task and send it off with CJobManager::Submit instead, or spawn a new thread; don't inherit from Cthread. Things like CAddonSystemSettings shouldn't be a thread.

FernetMenta · 2016-10-22T10:14:19Z

@phate89 using CThread may work but is not the best choice. What you really want to do is spawning an upgrade thread if required. Look into CJob and CJobManager. Those classes do what you need here.

EDIT : cross post :) @tamland had the same idea

phate89 · 2016-10-22T10:29:53Z

Thanks from the tip.. it'' quite nice ttt see we already have so many classes ready for several tasks.. I''l look into CJobManager and report back

phate89 · 2016-10-22T20:18:06Z

@tamland @FernetMenta I updated the pr using CJobManager::Submit. I hope I got it right this time :|

xbmc/Application.cpp

+    {
+      if (CDatabaseManager::GetInstance().m_bIsUpgrading)
+        CSplash::GetInstance().Show(std::string(i, ' ') + localizedStr + std::string(i, '.'));
+      if (notFinished)


xbmc/Application.cpp

+  {
+    for (int i = 1; i <= 3 && notFinished; ++i)
+    {
+      if (CDatabaseManager::GetInstance().m_bIsUpgrading)


FernetMenta · 2016-10-24T10:15:21Z

Looks ok from my pov. I can't say much about the upgrade itself

phate89 · 2016-10-24T11:51:44Z

jenkins build this please

tamland · 2016-10-24T15:11:15Z

Getting closer:) but you need to use std::future or events, and wait. With the sleeps the migration is up to 1 second slower than it need to be.

xbmc/addons/AddonSystemSettings.cpp

@@ -141,13 +141,15 @@ std::vector<std::string> CAddonSystemSettings::MigrateAddons(std::function<void(

  if (CSettings::GetInstance().GetInt(CSettings::SETTING_ADDONS_AUTOUPDATES) == AUTO_UPDATES_ON)
  {
-    onMigrate();
+    m_bIsMigratingAddons = true;


phate89 · 2016-10-24T15:26:57Z

@tamland
I know waiting is better but we also want to update the text every second to show some progress. It's something that can take several minutes so moving dots helps to see we're not stuck.

tamland · 2016-10-24T15:34:58Z

That's what it does. If you pass 1 second to the wait method it returns ready or timeout after 1 second. If it's ready before that it will be interrupted.

phate89 · 2016-10-24T17:25:10Z

@tamland I'm dumb of course you're right.. i updated it using our CEvent class

xbmc/Application.cpp

@@ -1123,8 +1124,22 @@ bool CApplication::Initialize()
  g_curlInterface.Unload();

  // initialize (and update as needed) our databases
-  CSplash::GetInstance().Show(g_localizeStrings.Get(24150));
-  CDatabaseManager::GetInstance().Initialize();
+  CEvent event;


phate89 · 2016-10-24T22:19:52Z

now it should be ok.. thanks @FernetMenta @tamland for all the help..
jenkins build this please

MartijnKaijser · 2016-10-25T16:42:18Z

win32 upgrade from v14 and all is fine. thx
jenkins build this please.

MartijnKaijser · 2016-10-26T07:49:13Z

@FernetMenta @tamland if all good please merge :)

FernetMenta · 2016-10-26T08:09:08Z

fine by me, @tamland your button

phate89 force-pushed the update_db branch from f1ca47f to 77f9b00 Compare October 19, 2016 15:27

DaveTBlake added the v17 Krypton label Oct 19, 2016

phil65 reviewed Oct 19, 2016

View reviewed changes

ksooo added the Type: Fix non-breaking change which fixes an issue label Oct 20, 2016

phate89 force-pushed the update_db branch from 77f9b00 to 29c0849 Compare October 20, 2016 10:16

phate89 force-pushed the update_db branch from 29c0849 to eec784c Compare October 20, 2016 10:41

FernetMenta reviewed Oct 20, 2016

View reviewed changes

phate89 closed this Oct 20, 2016

phate89 reopened this Oct 21, 2016

phate89 force-pushed the update_db branch 2 times, most recently from 95190fc to 2beef54 Compare October 22, 2016 09:32

phate89 force-pushed the update_db branch from 2beef54 to ba69721 Compare October 22, 2016 09:37

tamland reviewed Oct 22, 2016

View reviewed changes

xbmc/Application.cpp

if (CDatabaseManager::GetInstance().IsRunning())

Sleep(1000);

}

}

This comment was marked as spam.

Sign in to view

This comment was marked as spam.

Sign in to view

phate89 force-pushed the update_db branch from ba69721 to 786083b Compare October 22, 2016 20:16

phate89 force-pushed the update_db branch from 786083b to 48dc6c2 Compare October 22, 2016 23:30

DaveTBlake mentioned this pull request Oct 23, 2016

Forced rescan of music file tags for v17 #10759

Merged

FernetMenta reviewed Oct 23, 2016

View reviewed changes

xbmc/Application.cpp

{

if (CDatabaseManager::GetInstance().m_bIsUpgrading)

CSplash::GetInstance().Show(std::string(i, ' ') + localizedStr + std::string(i, '.'));

if (notFinished)

This comment was marked as spam.

Sign in to view

This comment was marked as spam.

Sign in to view

FernetMenta reviewed Oct 23, 2016

View reviewed changes

xbmc/Application.cpp

{

for (int i = 1; i <= 3 && notFinished; ++i)

{

if (CDatabaseManager::GetInstance().m_bIsUpgrading)

This comment was marked as spam.

Sign in to view

This comment was marked as spam.

Sign in to view

phate89 force-pushed the update_db branch from 06ccbe0 to a184181 Compare October 24, 2016 11:51

phate89 force-pushed the update_db branch 2 times, most recently from f9e72b7 to 438e7fc Compare October 24, 2016 13:41

tamland reviewed Oct 24, 2016

View reviewed changes

phate89 force-pushed the update_db branch from 438e7fc to 67c2cf9 Compare October 24, 2016 17:03

tamland reviewed Oct 24, 2016

View reviewed changes

phate89 force-pushed the update_db branch from 67c2cf9 to 5c5b448 Compare October 24, 2016 17:52

Change splash db upgrade logic

f344ca0

phate89 force-pushed the update_db branch from 5c5b448 to f344ca0 Compare October 25, 2016 15:24

MartijnKaijser added this to the Krypton 17.0-beta5 milestone Oct 25, 2016

tamland merged commit 7d97806 into xbmc:master Oct 27, 2016

phate89 deleted the update_db branch November 30, 2016 22:36

Change splash db upgrade logic #10731

Change splash db upgrade logic #10731

Conversation

phate89 commented Oct 19, 2016

Description

Motivation and Context

How Has This Been Tested?

Screenshots (if appropriate):

Types of change

Checklist:

phate89 commented Oct 19, 2016

DaveTBlake commented Oct 19, 2016

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

MilhouseVH commented Oct 19, 2016 • edited by DaveTBlake

DaveTBlake commented Oct 19, 2016

phate89 commented Oct 20, 2016

tamland commented Oct 20, 2016

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

Paxxi commented Oct 20, 2016

phate89 commented Oct 20, 2016 • edited

FernetMenta commented Oct 21, 2016

phate89 commented Oct 21, 2016 • edited

FernetMenta commented Oct 22, 2016

phate89 commented Oct 22, 2016

This comment was marked as spam.

This comment was marked as spam.

tamland commented Oct 22, 2016

FernetMenta commented Oct 22, 2016 • edited

phate89 commented Oct 22, 2016

phate89 commented Oct 22, 2016

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

FernetMenta commented Oct 24, 2016

phate89 commented Oct 24, 2016

tamland commented Oct 24, 2016

This comment was marked as spam.

phate89 commented Oct 24, 2016

tamland commented Oct 24, 2016

phate89 commented Oct 24, 2016

This comment was marked as spam.

phate89 commented Oct 24, 2016

MartijnKaijser commented Oct 25, 2016

MartijnKaijser commented Oct 26, 2016

FernetMenta commented Oct 26, 2016

MilhouseVH commented Oct 19, 2016 •

edited by DaveTBlake

phate89 commented Oct 20, 2016 •

edited

phate89 commented Oct 21, 2016 •

edited

FernetMenta commented Oct 22, 2016 •

edited