![]() ![]() ![]() > if (oa.mAppId != nsIScriptSecurityManager::NO_APP_ID || oa. DB4S uses a familiar spreadsheet-like interface, and complicated SQL commands do not have to be learned. DB4S is for users and developers who want to create, search, and edit databases. >+++ -60,16 +60,39 Scheme0Scope(DOMStorageCacheBridge* aCac DB Browser for SQLite (DB4S) is a high quality, visual, open source tool to create, design, and edit database files compatible with SQLite. >diff -git a/dom/storage/DOMStorageDBThread.cpp b/dom/storage/DOMStorageDBThread.cpp > Bug 1240238 - Full DOMStorage schema 1 scope update uniqueness, r=smaug try to create the index and on a failure just drop the database and start over with an empty oneĬomment on attachment 8711768 preserve uniqueness of scope to allow schema 1 -> schema 0 -> schema 1 jumps be fully handled Has user context already landed on m-c? If so, Vlad, have you been using it? I know about a certain problem that we loose uniqueness when origin attributes stored with schema 1 expands beyond what the "scope" column can store (which is only appid and inbrowser). I suspect we failed the update but the transaction didn't rollback correctly and left the schema broken. I'm not sure how that happened, since we don't allow writes to the database when we fail any part of the open and update process, half-updated tables should be threw away as all is happening in a transaction and is rolled back on a failure. that may happen only when the data were added to the database while there already were not a unique index (with bug 1165214 (=schema 1) it's origin_key_index) to fail the index creating is only possible when the scope+key pairs are not unique (I found 22 duplicates for ".: https:443" and "crosstab.MESSAGE_KEY") in Fx versions w/o bug 1165214 (=schema 0) we fail to create scope_key_index index -> hence the storage failures bubbling up to DOM the database has been apparently updated to schema 1, but the origin_key_index is missing from a reason I don't fully understand we don't try to create it again when already updated to schema 1 So, what's happening here (with the provided profile): A corrupt sqlite file with a new profile is hard to understand. What I find confusing about these errors is users should only see the firstrun flow when they create a new profile. Some of these errors occur several thousand times an hour, the main reported browsers are Fx 41->Fx 43. ![]() * SecurityError (This operation is insecure) Thanks Andy, we have been trying to track down the cause of this and related errors that are reported to FxA metrics. > Here's how the first run pages looks for me in that profile: (In reply to Andy McKay from comment #19) ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |