On the blog of the source of inspiration in open source, I saw this article: Open Source pure C # lightweight Database Engine: SharpHSQL 1.0.3.0.
I have used many similar database engines, such as sqlite and MINOSSE. Sqlite is inconvenient for pure c in web applications, and minnosse cannot run normally.
The main reason I am looking for such a database engine is that I want to replace ACCESS in web projects to reduce website construction costs and improve website performance.
First, I will compare SharpHsql with SQL Server 2000, in which Hsql leads the way in creating and deleting tables. The speed of data modification and query is basically the same. SharpHsql takes a little longer to query, and SharpHsql uses the cache technology.
When data is inserted, the data in the table increases and the speed slows down. When 10 thousand rows of data are inserted, the speed slows down. This speed is much slower than that of SQL server;
The data I used to test access is basically 1/10 of SQL server, so the SharpHsql speed is very advantageous compared with Access, and the overall performance is more than six times that of ACCESS.
SharpHsql databases have a total of 4 files *. data, *. cfg, *. log, and *. backup. You will know what to do with the suffix.
Let's take a look at the functions he supports. I will extract a piece of code here. You will understand it after reading the code.
Switch (sToken)
{
Case "SELECT ":
RResult = p. ProcessSelect ();
Break;
Case "INSERT ":
RResult = p. ProcessInsert ();
Break;
Case "UPDATE ":
RResult = p. ProcessUpdate ();
Break;
Case "DELETE ":
RResult = p. ProcessDelete ();
Break;
Case "ALTER ":
RResult = p. ProcessAlter ();
Break;
Case "CREATE ":
RResult = ProcessCreate (c, channel );
Script = true;
Break;
Case "DROP ":
RResult = ProcessDrop (c, channel );
Script = true;
Break;
Case "GRANT ":
RResult = ProcessGrantOrRevoke (c, channel, true );
Script = true;
Break;
Case "REVOKE ":
RResult = ProcessGrantOrRevoke (c, channel, false );
Script = true;
Break;
Case "CONNECT ":
RResult = ProcessConnect (c, channel );
Break;
Case "DISCONNECT ":
RResult = ProcessDisconnect (c, channel );
Break;
Case "SET ":
RResult = ProcessSet (c, channel );
Script = true;
Break;
Case "SCRIPT ":
RResult = ProcessScript (c, channel );
Break;
Case "COMMIT ":
RResult = ProcessCommit (c, channel );
Script = true;
Break;
Case "ROLLBACK ":
RResult = ProcessRollback (c, channel );
Script = true;
Break;
Case "SHUTDOWN ":
RResult = ProcessShutdown (c, channel );
Break;
Case "CHECKPOINT ":
RResult = ProcessCheckpoint (channel );
Break;
Case "CALL ":
RResult = p. ProcessCall ();
Break;
Case "SHOW ":
RResult = ProcessShow (c, channel );
Break;
Case "DECLARE ":
RResult = p. ProcessDeclare ();
Script = true;
Break;
Case ";":
Continue;
Default:
Throw Trace. Error (Trace. UnexpectedToken, Token en );
}
Then, when DataReader is used, its DataReader does not implement the indexer. It can only be implemented using DataReader [Columnname] And DataReader. GetInt32 (0. Note that the column names in DataReader [Columnname] must be in uppercase, and all SharpHsql statements in the system must be in uppercase;
DataReader [Columnname] is implemented using HashTable. DataReader. GetInt32 (0) is implemented using object.
When SharpHsqlDataAdapter is used, non-SharpHsqlDataAdapter is found. an error is reported in the form of Fill (dataset), such as SharpHsqlDataAdapter. fill (dataset, tablename); To solve this problem, you only need to delete the SharpHsqlDataAdapter. the Fill Method in cs is enough. In fact, it does not need to be rewritten. Public override int Fill (DataSet dataSet)
{
Return base. Fill (dataSet );
}
The problems I have found are just like this.
By the way, his log4net system actually occupies space. I will remove it!
In general, the database engine is still relatively well-developed, but the driver write is relatively sloppy, and the data object names should not all be replaced with uppercase letters.
