Mono 2.4.3 doesn't like this when running nunit, with nunit throwing
AssetTests`2 : System.MemberAccessException : Cannot create an instance of OpenSim.Data.Tests.AssetTests`2[TConn,TAssetData] because Type.ContainsGenericParameters is true.
and similar. Mono 2.10.8.1 does not have this issue.
So will wait until min version of mono bumps before restoring.
Aims to address http://opensimulator.org/mantis/view.php?id=6355
As part of this work, an incomplete IXGroupsData was added which currently only allows store/fetch/delete of group records
(i.e. no membership data etc)
This is subject to change and currently only an in-memory storage implementation exists for regression test purposes.
This is done in the way that one would expect (e.g. moving a folder increments version number on both source and destination parent folders).
This should hopefully improve viewer reuse of its cached inventory information.
Currently MySQL only but will be implement for SQLite/MSSQL if there are no issues.
This is to stop viewer inventory cache version numbers becoming out of sync with grid stored numbers when viewer performs these actions.
If there are no problems with these changes, they will be propogated to SQLite (and MSSQL if that's simple enough).
May also need to do the same on folder store/create/delete and maybe propogate version increments up the folder hierarchy, but that requires investigation.
This was not necessary with the nant build but is for xbuild, since xbuild includes associated project DLLs
And so picks up OpenMetaverse.dll even though it isn't referenced by OpenSim.Data.Null.dll
This prevented more than one additional ordinary folder from being created in the base "My Inventory" user folder.
Added regression test for this case.
Switched tests to use XInventoryService with mostly implemented TestXInventoryDataPlugin rather than InventoryService
Disabled TestLoadIarV0_1SameNameCreator() since this has not been working for a very long time (ever since XInventoryService) started being used
since it doesnt' preserve creator data in the same way as InventoryService did and so effectively lost the OSPAs.
However, nobody noticed/complained about this issue and OSPAs have been superseded by HG like creator information via the --home save oar/iar switch.
This is always done later on in SceneGraph.AddSceneObject() if the call hasn't failed due to sanity checks.
There's no other purpose for this method to exist and it's dangerous/pointless to call in other conditions.
Previously NULL fields were converted to an empty string due to the use of ToString(). But if the field was an Int (e.g., "locZ"), then the subsequent attempt to convert an empty string to an int caused an exception. Now the field is null so we don't try to convert it, so there's no exception.
This may no longer be an issue with better locale enforcement in OpenSimulator but it doesn't hurt to have this patch.
http://opensimulator.org/mantis/view.php?id=4680
Thanks to controlbreak for this
Rename IFriendsModule.GetFriendPerms() -> GetRightsGrantedByFriend() to be more self-documenting and consistent with friends module terminology.
Add some method doc.
The asset data plugin now implements IXAssetData rather than IAssetData so the ordinary AssetService should no longer pick it up.
This replaces the changes in 92b1ade. There is no longer any need to adjust your StandaloneCommon.ini/Robust.ini/Robust.HG.ini files.
This may explain very recent issues in the last few weeks where textures have been disappearing or turning white (as they were going to different places).
Unfortunately, you will need to rollback to an earlier database backup or reupload the textures.
.NET 4.0 added the method Stream.CopyTo(stream, bufferSize). For .NET 3.5
and before, WebUtil defined an extension method for Stream with the signature
Stream.CopyTo(stream, maxBytesToCopy). The meaning of the second parameter
is different in the two forms and depending on which compiler and/or
runtime you use, you could get one form or the other. Crashes ensue.
This change renames the WebUtil stream copy method to something that
cannot be confused with the new CopyTo method defined in .NET 4.0.
This should not currently be used in any circumstances except for experimentation.
Database tables used by this plugin can still change at any time with no migration path.
Some successful collision attacks have been carried out on sha1 with speculation that more are possible.
http://en.wikipedia.org/wiki/Cryptographic_hash_function#Cryptographic_hash_algorithms
No successful attacks have been shown on sha256, which makes it less likely that anybody will be able to engineer an asset hash collision in the future.
Tradeoff is more storage required for hashes, and more cpu to hash, though this is neglible compared to db operations and network access.
Whether this is worthwhile is debatable since here we are not transmitting data over a network
In addition, jpeg2000 (the biggest data hog) is already a compressed image format.
May not remain.
This is an extremely crude implemenation which almost works by accident. Nevertheless it does work.
It can be tested with the instructions at http://opensimulator.org/wiki/Feature_Proposals/Deduplicating_Asset_Service#Testing
It does not interact at all with the existing asset service or any data stored there.
This code is subject to change without notice and should not be used for anything other than gawking.
MySQL and MSSQL have it as Sandbox, sqlite as sandbox.
In various different places in every plugin the wrong casing is used...
Consistency, who needs it? Or one day sqlite can change to Sandbox.
Telehub commits! They will eat your babies and corrupt your database while
they munch. DO NOT use anything from the first Telehub commit to this one.
FIRST GOOD COMMIT is the one FOLLOWING this one. You have been warned.
This is damage control es EstateSettings is not the place this can be put.
EstateSettings is nt unique to a region and therefore would introduce
a hard limit of one telehub per estate, completely shutting off the
option of having SL style telehubs, e.g. one per region. Whole
estate teleport routing can still be implemented id desiresd, this
way all options are open while the other way most options get closed
off.
Telehub settings now persist to the database and are saved across sim restarts. So-far this only works on MySQL. this is a work in progress, teleport routing is not yet implemented.
This required an option to be added to NullRegionData via ConnectionString for it to act as a non-static instance, so that regression tests (which only load this class once) don't get hopeless confused and complex to compensate.
Normal standalone operation unaffected.
These are just the result of an attempt to canonicalize received messages - it's not important that we constantly log them.
Also finally get the deregister grid service message working properly
UpdateFlag is now referenced/used only within SOP and SOG. Outsiders are
using ScheduleFullUpdate, ScheduleTerseUpdate or ClearUpdateSchedule on
SOP consistently now. Also started working toward eliminating those
calls to ScheduleFullUpdate, ScheduleTerseUpdate or ClearUpdateSchedule
from outside SOP in favor of just setting properties on SOP and let SOP
decide if an update should be scheduled. This consolidates the update
policy within SOP and the client rather than everywhere that makes
changes to SOP. Some places forget to call update while others call it
multiple times, "just to be sure".
UpdateFlag and Schedule*Update will both be made private shortly.
UpdateFlag is intended to be transient and internal to SOP so it has
been removed from XML serializer for SOPs.
Using the RemoteAdmin API to close then recreate a region would fail if the estate name had not changed. If the estate name /was/ changed then the existing estate would be renamed rather than a new one being created. The problem really arose from a lack of distinction in the data storage layer between creating new estates and loading existing ones.
Leaving them at UUID.Zero meant that when a viewer 2 logged into a region that had been freshly created, it received UUID.Zero for these textures, and hence display the land as plain white.
On a simulator restart, the problem would go away since when the database adapators loaded the new region settings, RegionSettings itself has code to use default textures instead of UUID.Zero.
This commit resolves the problem by saving the default texture UUIDs instead of Zero.
However, we currently have to do this in a roundabout way by resaving once the RegionSettings have been created by the database for the first time. This needless complexity should be addressed.
This change will also have the effect of replacing any existing UUID.Zero terrain textures with the default ones.
However, this shouldn't have any effect since the UUID.Zeros were already being replaced in memory with those same UUIDs.
Using MyISAM proves vastly faster for persisting scene objects.
For instance, a scene object that took 9 seconds to persist before now takes 1. This also improves the experience of loading large OARs.
We don't use any of the transactional features of InnoDB.
The only thing that may have an impact is that InnoDB does row locking on inserts while MyISAM does table locking.
However, field reports say there is no noticeable difference.
Revert "Changed Friends table to have 165-sized varchars on PrincipalID and FriendID. The reason for this number is the following: there is a combined key of these 2 fields; apparently MySql can't handle keys larger than 1000 bytes; when the table is created with utf8 encoding, this combined key is bigger than 1000 bytes, and the migration fails. WARNING: this is not a new migration! People who have gone through this migration and failed should update the sizes of these fields manually."
This reverts commit 3fa54a156a.
If these links are not deleted, then they will build up in the player's inventory until they can no longer log in.
Accidental deletion of links due to bugs or other causes is potentially inconvenient but on a par with items being
accidentally moved. When a link is deleted, the target of the link is never touched.
This is a general solution that accounts for the use of links anywhere in the user's inventory.
If these links are not deleted, then they will build up in the player's inventory until they can no longer log in.
Accidental deletion of links due to bugs or other causes is potentially inconvenient but on a par with items being
accidentally moved. When a link is deleted, the target of the link is never touched.
This is a general solution that accounts for the use of links anywhere in the user's inventory.
This applies to adding/removing estate users, groups, managers and bans.
This is the application of the AllEstates_0.5.patch from http://opensimulator.org/mantis/view.php?id=5420
Thanks very much, Snoopy!
Run alter table estate_settings add column AllowLandmark tinyint not null default 1, add column AllowParcelChanges tinyint not null default 1, add column AllowSetHome tinyint not null default 1;
to make this work.
The in-memory storage of region data that is used by default with SQLite now handles wildcards in region names in the same way as SQL queries do with other databases.
This matches the behaviour for other deletions and stops this operation failing on Windows .NET (though mono is fine with the duplication).
Probably a holdover from copying code from the older legacy adaptor.
Resolves http://opensimulator.org/mantis/view.php?id=5361
set in the target's UserFlags and the impersonator must have UserLevel 200
or above. The user can then log in using the target's name and their own
password.
New option added to save oar -profile=url. Migration on RegionStore making CreatorID be 255 chars.
Moved Handling of user UUID -> name requests to a new module UserManagement/UserManagementModule.
midday defaults when no specific LightShare profile is set.
This prevents LightShare info being send out when the region has no LightShare
profile, allowing normal day/night cycles to happen.
This will cause visual params to be persisted along with worn items. With
this, alpha and tattoo laters will be saved. Multiple layers MAY work, but
not tested because I don't use Viewer 2.
* Hopefully this will enable mac people to use SQLite again
* Someone cooler then me will need to figure out how to define CSharpSqlite with Mono so the conditional will pick up.
* There's also an error that occurs when you first load OpenSimulator under CSharpSQLite that goes away after.
* The databases are not byte for byte compatible. Mono.Data.Sqlite is able to read CSharpSqlite made databases but not the reverse way.
* Use a conditional define to determine whether we're using CSharpSqlite or Mono.Data.Sqlite
#if CSharpSqlite
using Community.CsharpSqlite.Sqlite;
#else
using Mono.Data.Sqlite;
#endif
* Hopefully, this will restore sqlite functionality on a Mac. In visual studio, you can edit the OpenSim.Data.SQLite project, go to the Build tab and enter CSharpSqlite in the box. I'm not sure how to define CSharpSqlite in Mono, someone better at it then me will have to take the job of figuring out the best way to define it in Mono.
When running for the first time, people see migration failures because of sql statements that are trying to move data from old tables (e.g. users). The amended text attempts to calm their nerves.
* Removed StorageManager
* CONFIG CHANGE: There are no more database settings in OpenSim.ini. Check the config-include configuration files for region store and estate store database settings
Previously, Scene.Inventory.DeRezObjects() forced the persistence of prims before deletion.
This is necessary so that freshly delinked prims can be deleted (otherwise they remain as parts of their old group and reappear on server restart).
However, DeRezObjects() deleted to user inventory, which is not required by llDie() or direct region module unlink and deletion.
Therefore, forced persistence has been pushed down into Scene.UnlinkSceneObject() to be more general, this is still on the DeRezObjects() path.
Uncommented TestDelinkPersistence() since this now passes.
Tests required considerable elaboration of MockRegionDataPlugin to reflect underlying storing of parts.
Previously, Scene.Inventory.DeRezObjects() forced the persistence of prims before deletion.
This is necessary so that freshly delinked prims can be deleted (otherwise they remain as parts of their old group and reappear on server restart).
However, DeRezObjects() deleted to user inventory, which is required by llDie() or direct region module unlink and deletion.
Therefore, forced persistence has been pushed down into Scene.UnlinkSceneObject() to be more general, this is still on the DeRezObjects() path.
Uncommented TestDelinkPersistence() since this now passes.
Tests required considerable elaboration of MockRegionDataPlugin to reflect underlying storing of parts.
This shows that panda is still not running the sqlite database tests since this change fixes the test failure in that area when "nant test" is executed locally
using an OSD serialization rather than auto forces serialization code to be placed in OpenSim.Framework
this makes the media texture raw data staging post in PrimitiveBaseShape redundant, now we just directly call the code in PrimitiveBaseShape.MediaList itself
This is better modularity. It also allows MoapModule to be replaced with some other media module that may behave completely differently in the future.
Remaining non-modularity:
PrimitiveBaseShape needs explicit Media and MediaRaw fields. MediaRaw is required in order to shuttle the pre-serialization data back and forth from the database layer.
The database also needs to know about MediaRaw though not about Media.
IMO, it would be extremely nice to remove these hard codings but this is a bridge too far at the present time.
THIS WILL BREAK EXISTING MEDIA TEXTURE PERSISTENCE. Please delete your existing sqlite databases if you are experimenting with this branch.
This wrapper will make it easier to maintain compatibility if the media texture data evolves.
This will also make it easier to store non-sl media texture data.
Change the slam bit from 3 to 4. Assume the old slam bit is always set.
The new slam bit is a "changed owner" bit, correcting a bug where an item
passed from the creator to another with less than full perms, then back (sale
test) would arrive back full perm. Lots of in-code docs.
Change the slam bit from 3 to 4. Assume the old slam bit is always set.
The new slam bit is a "changed owner" bit, correcting a bug where an item
passed from the creator to another with less than full perms, then back (sale
test) would arrive back full perm. Lots of in-code docs.
this appears to be okay even though we reuse it between runs without deleting it first.
size of the database appears not to be changing though that could be deceptive
The base test class now tries to connect to DB, ignores all tests in the
class if unable to.
Also m_log changed to instance field (which in this case shouldn't cause
any problems), to avoid having to define it separately in each derived
class. Here I touched things that I don't understand well (using log4net),
so please review this commit.
There was a whole bunch of these SQL files, all empty and apparently
unused. Removing them is just a clean-up, if anybody has a reason for these
files to be there, feel free to revert.
Besides, AssetData is slightly optimized to StoreAsset in one request
("IF EXISTS() UPDATE ... ELSE INSERT ...")
The main change in the MS SQL Inventory implem. is that it now return
empty list (or whatever) when called with UUID.Zero, which is consistent
with how the code for other DBs work.
I did no changes at all in XInventory, as there is no test set for them.
ok, so the estate stores now want their own migration files, but as it
happened the SQL definition were inside the Region migrations.
It seems better/cleaner to keep each 'store' separately updatable.
WARNING: any editing in the middle of the migration scripts (as opposite
to just appending to them) has the potential of messing up updates of
existing databases. As far as I can see, this one is (probably) safe,
the worst that could happen is the EstateStore migration silently fail
if the estate the tables are already there.
The tests have been modified to work under NUnit 2.4.6
(the one currently used in the project).
They will also work with NUnit 2.5+ as is, but will look better
if you #define NUNIT25 for them.
NOTE that this INI file is currently loaded as a embedded RESOURCE, which is weird and has a
disadvantage of having to rebuild the Tests whenever the conn strings are changed.
The only reason is that I couldn't figure out a reliable way to put this INI into the correct
dir at runtime. If somebody can do it, that would be cool.
AssetTests: The name has changed to reflect the fact it is no longer a base class,
but the complete asset test for all supported databases.
The test can also check storing of CreatorID, but the feature is
disabled at this commit!
The problem was that some tests relied on prior tests
to leave the DB in a particular state, but the test class
cleared the DB every time. The affected tests have been
merged into one to remove the dependencies.
tested on all 3 Dbs, all tests green.
Fixed unfinished SQL in FetchAssetMetadataSet, fixed SQL
in UpdateAsset (must not modify ID).
NOT tested! But apparently shouldn't work worse than the previous version,
esp. the FetchMetadata thing.
These are some generic classes that simplify writing tests
for any of the data connectors and databases. Among other
things, configuring the connection strings is done once,
in a separate resource file.
Tests based on the new BasicDataServiceTest class require
NUnit 2.5 or better.
Contains a migration.
SQLite: May contain nuts.
The SQLite migration copies the entire asset table. Be prepared for quite a wait. Don't interrupt it. Back up your assets db.
BasicAssetTest checks CreatorID storage, new test for weird CreatorID
(now also checks that non-GUID or empty CreatorID gets stored correctly)
Signed-off-by: Melanie <melanie@t-data.com>
The asset type wasn't in the list of "DontScramble" fields,
so the test assets were stored with randomized type, which
caused exception on reading them.
Also the scrambler was moved from local var to the class level,
so it could be used in the new tests I've added (see the next commit).
Throwing an Ex. with SQL command in the message looks weird,
this is a bit better, but I'm still not sure if that's the
proper way to handle. Also, there is a catch one level up,
so is this one necessary?
Change migration error reporting to not truncate the statement when
reporting. It's a bit messier than the old error reporting, but at least
one gets an idea of what could be wrong again. And things look a lot
neater now.
Melanie