Add new file doc/wal2.md to this branch.

FossilOrigin-Name: b495dce153f7f886f3dba09593f29ce2375718bf6508f2cfffd1af8071a995ae

doc/wal2.md

@@ -0,0 +1,98 @@
+
+Wal2 Mode Notes
+===============
+
+## Activating/Deactivating Wal2 Mode
+
+"Wal2" mode is very similar to "wal" mode. To change a database to wal2 mode,
+use the command:
+
+>
+     PRAGMA journal_mode = wal2;
+
+It is not possible to change a database directly from "wal" mode to "wal2"
+mode. Instead, it must first be changed to rollback mode. So, to change a wal
+mode database to wal2 mode, the following two commands may be used:
+
+>
+     PRAGMA journal_mode = delete;
+     PRAGMA journal_mode = wal2;
+
+A database in wal2 mode may only be accessed by versions of SQLite compiled
+from this branch. Attempting to use any other version of SQLite results in an
+SQLITE_NOTADB error. A wal2 mode database may be changed back to rollback mode
+(making it accessible by all versions of SQLite) using:
+
+>
+     PRAGMA journal_mode = delete;
+
+## The Advantage of Wal2 Mode
+
+In legacy wal mode, when a writer writes data to the database, it doesn't
+modify the database file directly. Instead, it appends new data to the
+"<database>-wal" file. Readers read data from both the original database
+file and the "<database>-wal" file. At some point, data is copied from the
+"<database>-wal" file into the database file, after which the wal file can
+be deleted or overwritten. Copying data from the wal file into the database
+file is called a "checkpoint", and may be done explictly (either by "PRAGMA
+wal_checkpoint" or sqlite3_wal_checkpoint_v2()), or
+automatically (by configuring "PRAGMA wal_autocheckpoint" - this is the
+default).
+
+Checkpointers do not block writers, and writers do not block checkpointers.
+However, if a writer writes to the database while a checkpoint is ongoing,
+then the new data is appended to the end of the wal file. This means that,
+even following the checkpoint, the wal file cannot be overwritten or deleted,
+and so all subsequent transactions must also be appended to the wal file. The
+work of the checkpointer is not wasted - SQLite remembers which parts of the
+wal file have already been copied into the db file so that the next checkpoint
+does not have to do so again - but it does mean that the wal file may grow
+indefinitely if the checkpointer never gets a chance to finish without a
+writer appending to the wal file. There are also circumstances in which
+long-running readers may prevent a checkpointer from checkpointing the entire
+wal file - also causing the wal file to grow indefinitely in a busy system.
+
+Wal2 mode does not have this problem. In wal2 mode, wal files do not grow
+indefinitely even if the checkpointer never has a chance to finish
+uninterrupted.
+
+In wal2 mode, the system uses two wal files instead of one. The files are named
+"<database>-wal" and "<database>-wal2", where "<database>" is of
+course the name of the database file. When data is written to the database, the
+writer begins by appending the new data to the first wal file. Once the first
+wal file has grown large enough, writers switch to appending data to the second
+wal file. At this point the first wal file can be checkpointed (after which it
+can be overwritten). Then, once the second wal file has grown large enough and
+the first wal file has been checkpointed, writers switch back to the first wal
+file. And so on.
+
+## Application Programming
+
+From the point of view of the user, the main differences between wal and 
+wal2 mode are to do with checkpointing:
+
+  * In wal mode, a checkpoint may be attempted at any time. In wal2 
+    mode, the checkpointer has to wait until writers have switched 
+    to the "other" wal file before a checkpoint can take place.
+
+  * In wal mode, the wal-hook (callback registered using
+    sqlite3_wal_hook()) is invoked after a transaction is committed
+    with the total number of pages in the wal file as an argument. In wal2
+    mode, the argument is either the total number of uncheckpointed pages in
+    both wal files, or - if the "other" wal file is empty or already
+    checkpointed - 0.
+
+Clients are recommended to use the same strategies for checkpointing wal2 mode
+databases as for wal databases - by registering a wal-hook using
+sqlite3_wal_hook() and attempting a checkpoint when the parameter
+exceeds a certain threshold.
+
+However, it should be noted that although the wal-hook is invoked after each
+transaction is committed to disk and database locks released, it is still
+invoked from within the sqlite3_step() call used to execute the "COMMIT"
+command. In BEGIN CONCURRENT systems, where the "COMMIT" is often protected by
+an application mutex, this may reduce concurrency. In such systems, instead of
+executing a checkpoint from within the wal-hook, a thread might defer this
+action until after the application mutex has been released.
+
+

manifest

@@ -1,5 +1,5 @@
-C Reinstate\sassert()\sstatements\sin\sos_unix.c\sthat\swere\sremoved\sto\sallow\swal-mode\nSHARED\slocks\sto\sbe\staken\sover\smore\sthan\sone\slocking\sslot\s(this\sbranch\sno\nlonger\sdoes\sthat,\sso\sthe\sassert()\sstatements\scan\sgo\sback\sin).
-D 2019-01-02T17:00:00.197
+C Add\snew\sfile\sdoc/wal2.md\sto\sthis\sbranch.
+D 2019-01-11T14:58:03.764
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F Makefile.in d8b254f8bb81bab43c340d70d17dc3babab40fcc8a348c8255881f780a45fee6
@@ -40,6 +40,7 @@ F doc/F2FS.txt c1d4a0ae9711cfe0e1d8b019d154f1c29e0d3abfe820787ba1e9ed7691160fcd
 F doc/lemon.html 24956ab2995e55fe171e55bdd04f22b553957dc8bb43501dbb9311e30187e0d3
 F doc/pager-invariants.txt 27fed9a70ddad2088750c4a2b493b63853da2710
 F doc/vfs-shm.txt e101f27ea02a8387ce46a05be2b1a902a021d37a
+F doc/wal2.md a807405a05e19a4945c5905a9ffa0fe45b8560dd7572461192501f565c19cdb5
 F ext/README.md fd5f78013b0a2bc6f0067afb19e6ad040e89a10179b4f6f03eee58fac5f169bd
 F ext/async/README.txt e12275968f6fde133a80e04387d0e839b0c51f91
 F ext/async/sqlite3async.c 0f3070cc3f5ede78f2b9361fb3b629ce200d7d74
@@ -1806,7 +1807,7 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 87ef68f9174b5e21fa2fb4f2fa23e0ecd6a0ea854422c8aa74cd7306a1ed8427
-R 81c151bc78b917b5a505ba7d0041b59f
+P 8445fb6d5dae98a0129514a10a15f08412bac5a1d8114b12e592c744ea2bc4c9
+R 4b991801ac5f384be1f414fb71a05937
 U dan
-Z 69b1e094c88bc1f9442e8f44aee16f43
+Z 736399187a1b654bff250254a2a37532

manifest.uuid

@@ -1 +1 @@
-8445fb6d5dae98a0129514a10a15f08412bac5a1d8114b12e592c744ea2bc4c9
+b495dce153f7f886f3dba09593f29ce2375718bf6508f2cfffd1af8071a995ae