Difference between revisions of "Projects/PIM/MS Windows/SQLite Folder Indices"

Latest revision as of 10:35, 16 May 2008

There are issues with locking index files for KMail folders and mmap()/munmap() operations on Windows. Therefore, SQLite-based indices are in development. This page presents detailed development notes for this task.

Design and implementation started by jstaniek 11:35, 23 April 2008 (CEST)

[edit] Installation

Update (7 May 2008): commit 805075 merged changes related to the SQLite mode for KMail indices from /branches/work/kmail-nommap (r799390..804487) and /branches/work/kdepim-nommap/kmail (r804484..804960) back to trunk.

Thus, it is enough to use kdepim trunk now. Users of kdepim-nommap should execute emerge --unmerge kdepim-nommap and then emerge kdepim.

Old instructions follow: branches/work/kdepim-nommap branch has been created for SQLite mode, being synced with kdepim trunk. The only difference is kmail/ subdirectory. To build it on Windows:

The last command will download the source code from the branch, with altered kmail/ source code, compile it and install. All the resulting files have the same names as in regular kdepim trunk. sqlite package will be also installed as a hard dependency of the kdepim-nommap package.

[edit] Introduction

• we call the new implementation SQLite mode for short and the old implementation mmap mode
• SQLite 3.5.4 is used, as provided by emerge sqlite module; we should not allow using much older versions of sqlite, e.g. 3.1 because of file format differences
• we are using one .index.db "index file" per folder (on request, could be possible implement it per-account)
• kmailprivate links to sqlite library for SQLite mode, and KMAIL_SQLITE_INDEX is defined to enable #ifdef'd code
• kmfolderindex_sqlite.cpp is created and edited as a copy of kmfolderindex.cpp; kmfolderindex.cpp #includes kmfolderindex_sqlite.cpp for SQLite mode and then skips its own code completely
• kmfolderindex_common.cpp is always included by kmfolderindex.cpp; implements KMFolderIndex::openInternal() and KMFolderIndex::createInternal()
• kmfolderindex.h is a common header for both kmfolderindex*.cpp implementations

[edit] KMFolderIndex

api docs

• 2008-04-23..25
  • mIndexId unused - removed as well as serialIndexId()
  • indexLocation(): added .db suffix to indicate the index is sqlite-based, implementation moved to FolderStorage (before FolderStorage only had it as abstract method)
  • INDEX_VERSION is written and checked using 'PRAGMA user_version = <integer>' command [1]
  • we do not use temporary filenames, e.g. in writeIndex(): SQLite takes care about safe storage
  • updateIndex(): no changes, we're changing implementation of KMMsgBase::syncIndexString() and writeIndex() instead
  • added openDatabase( int mode ) for SQLite mode
  • readIndex() implemented for SQLite mode - uses SELECT command on messages table
  • readIndexHeader() uses "PRAGMA user_version" sqlite command
  • writeIndex() implemented for SQLite mode - uses INSERT command on messages table, within transaction
  • common code from {KMFolderMbox|KMFolderMaildir}::open( const char * ) moved to KMFolderIndex::openInternal()
  • common code from {KMFolderMbox|KMFolderMaildir}::create() moved to KMFolderIndex::createInternal()
• 2008-05-01..02
  • implemented updateIndex() (see the table) and added extended version of writeMessages() with WriteMessagesMode modes
  • writeMessages(): depending on the mode (WriteMessagesMode) the sql statement "INSERT INTO messages(data) VALUES(?)" or "INSERT OR REPLACE INTO messages(id, data) VALUES (?, ?)", thus unique IDs are created when new yet unsored messages are saved, for later use
  • readIndex() now calls "SELECT id, data FROM messages"
  • 'messages' table creation is now: "CREATE TABLE messages( id INTEGER PRIMARY KEY AUTOINCREMENT, data BLOB )" - added autoincremented 'id' PK column

[edit] FolderStorage

• as implementation of KMFolderIndex::indexLocation() is moved to FolderStorage (before FolderStorage only had it as abstract method); FolderStorage::idsLocation() and FolderStorage::sortedLocation() are added to avoid performing the math like mFolder->indexLocation() + ".sorted"
• added QString location(const QString& suffix) - returns full path to .index, .ids or .sorted file (depending on the suffix)

[edit] KMMsgBase, KMMsgInfo

api docs

• 2008-04-23..25
  • added char* mData for SQLite mode only (and a getter/setter)
  • getStringPart() and getLongPart() share code between modes now: only cosmetic changes applied
• 2008-05-02
  • 'sqlite_int64 dbId' member added to KMMsgBase; KMMsgBase and KMMsgInfo ctors in SQLite mode now differ from mmap mode
  • KMMsgBase::syncIndexString() completely removed in SQLite mode because we want to keep all SQLite-specific operations in kmfolderindex_sqlite.cpp; KMFolderIndex::updateIndex() does this now.

[edit] KMFolderDir

• reload(): skip *.index.db files

[edit] Other classes

• 2008-05-02
  • MessageProperty: use ConstIterators for QMap structures to avoid double lookups

[edit] Status of porting to SQLite

[edit] Important Commits

• 802868 - partially working implementation (Wed Apr 30 22:20:39 2008 UTC)
• 803451 - 'messages' table creation is now: "CREATE TABLE messages( id INTEGER PRIMARY KEY AUTOINCREMENT, data BLOB )" - added autoincremented 'id' PK column; updated all the code for this design; the value of ID column is now used for updating the dirty records (Fri May 2 22:12:35 2008 UTC)
• 804186 - KMFolderIndex::updateIndex() - fix result checking of writeMessages() -fetching emails from (d)imap seems to work smoothly now (Mon May 5 09:51:46 2008 UTC)
• 805075 - changes related to the SQLite mode for KMail indices from /branches/work/kmail-nommap (r799390..804487) /branches/work/kdepim-nommap/kmail (r804484..804960) merged into trunk
  • See the page devoted to validation of the merege and the results.

[edit] Open Questions

• should we port .sorted and .ids files to sqlite too? (possibly to the same .db files)