[h2] Problem with TRACE_LEVEL_FILE option

Hi, I'm trying to turn off the *.trace.db file but it seems to be not working. I have tried (like explained here ) to set the TRACE_LEVEL_FILE option to 0 by adding ' ;TRACE_LEVEL_FILE=0' to the database URI or by calling the SQL

Re: [h2] Re: DB growing just by reading LOBs

On Thursday, 7 December 2017 15:13:40 UTC+1, Noel Grandin wrote: > > > Of course. Under most conditions (ie. with setAutoCommit==true), the > transaction is complete and closed before the > ResultSet is even returned to the client. > This is how JDBC works on most databases. > A read-only

Re: [h2] Re: DB growing just by reading LOBs

On Thursday, 7 December 2017 10:33:36 UTC+1, Noel Grandin wrote: > > > Backwards compatibility is important to us, and the H2 code has worked > this way since I started working on it. > I totally agree backward compatibility is important. That is why I opted for a (connection/database)

Re: [h2] Re: DB growing just by reading LOBs

On Thursday, 7 December 2017 12:24:10 UTC+1, Noel Grandin wrote: > > > That is doable, but has nasty performance surprises if you, for example, > do SELECT * on a table with a LOB column. If > the LOB is big enough, it gets buffered to disk on the client side, and > your SELECT performance is

Re: [h2] Re: DB growing just by reading LOBs

On 2017/12/07 9:42 AM, Silvio wrote: Why not transfer all LOBs to the client at once during transaction closing when it does such a nonsensical thing as reading them after closing the transaction they where acquired in? Backwards compatibility is important to us, and the H2 code has

Re: [h2] Re: DB growing just by reading LOBs

On 2017/12/07 12:20 PM, Silvio wrote: When all open LOBs are transferred to the client side of the connection when the transaction is closed they can still be read by the application afterwards. The only noticeable difference in behavior would be a potential performance hit and memory

Re: [h2] Usage of 'classpath:/filename.csv' - SQLSyntax and IOExceptions

Hi,I'm using the syntax in the documentation, as mentioned: this is at http://www.h2database.com/html/functions.html.Under the section headed 'CSVREAD" it states: "Instead of a file, an URL may be used, for example jar:file:///c:/temp/example.zip!/org/example/nested.csv. To read a stream from

[h2] Re: Usage of 'classpath:/filename.csv' - SQLSyntax and IOExceptions

The following code seems to work; so it's possible to use a URL, but not the documented classpath: protocol. I haven't tried reconfiguring this to run from a jar as yet, which is the object of the exercise. String url = getClass().getClassLoader().getResource(

Re: [h2] Re: DB growing just by reading LOBs

On 2017/12/07 1:37 PM, Silvio wrote: Does that mean that you allow reading the resultset AFTER the transaction completes? I was assuming the transaction Of course. Under most conditions (ie. with setAutoCommit==true), the transaction is complete and closed before the ResultSet is even

[h2] Add NATURAL (LEFT | RIGHT ) JOIN support.

How difficult would it be to add NATURAL (LEFT|RIGHT) JOIN support? By looking at the class Parser.java it looks like it can be done at the level of Parser by essentially "cut and paste" some code from two if branches if (readIf("RIGHT")) { else if (readIf("LEFT")) to else if

Re: [h2] Add NATURAL (LEFT | RIGHT ) JOIN support.

I think mean NATURAL LEFT OUTER and NATURAL RIGHT OUTER ? Because the default (AFAIK) is an INNER join. In which case those are the Parser changes, yes, but you'd need to set up the join condition properly as well. ​ -- You received this message because you are subscribed to the Google Groups

Re: [h2] Re: DB growing just by reading LOBs

By the time getResultSet() returns a ResultSet object, a transaction is normally completed, and all rows are buffered on the client side. So there is no copy-on-write type logic necessary for normal rows. However Blobs are different. For LOBs we just send the client a kind of pointer in the

Re: [h2] Add NATURAL (LEFT | RIGHT ) JOIN support.

Yes, OUTER is assumed. I would like to go ahead and try to implement the changes. Join conditions are already set up in else if (readIf("NATURAL")) branch and I assume they are the same for OUTER and INNER NATURAL joins. BTW, what is the current protocol for submitting patches? This link

Re: [h2] Add NATURAL (LEFT | RIGHT ) JOIN support.

please submit a pull request on github​ -- You received this message because you are subscribed to the Google Groups "H2 Database" group. To unsubscribe from this group and stop receiving emails from it, send an email to h2-database+unsubscr...@googlegroups.com. To post to this group, send