Subject Re: Question about URL and J2EE and a very little suggestion
Author rrokytskyy
> 1. Is there any reason why the connection URL looks like this?
> drivername:hostname/portname:/database
>
> Usually this looks like
> drivername://hostname:port/database
>
> is it a bug or is it a feature? [:)]

Feature :). The connection URL to the database in the Firebird API
looks exactly hostname/port:/database. So, we tried to be consistent
with Firebird. If Oracle can have their own URL representation, why
cannot we have our own? :)

> 2. I could not use the driver without installing J2EE. Actually I
> really don't need it, so will it be required in the future?

Strange... I do not have J2EE installed, but I can build it without
any problems. Yes, driver uses some J2EE interfaces, but they all are
in src/lib directory. What problems did you have building the driver
without J2EE installed?

> 3. I've tried to compile the driver with JDK 1.4, but I could not.
> Not a big thing, I've made a clean a JDK 1.3 install and it worked,
> I had to only replace the JAVA_HOME variable. My suggestion is to
> add %JAVA_HOME%/bin/ to last line in build.bat so this way it will
> always use the right java version, not that one that is on the path.

One of the developers, Alejandro Alberola Arias, takes care about JDK
1.4. So, soon we should have clean build with JDK 1.4 too (maybe, it
is already there, try updating the sources from CVS).

> so the last line should look like:
>
> %JAVA_HOME%/bin/java -classpath %LOCALCLASSPATH%
org.apache.tools.ant.Main %ANT_CMD_LINE_ARGS%

I checked couple of open-source projects, and all they
use "java ...", not "%JAVA_HOME%/bin/java". I would keep this as it
is, but if people here vote to change it, I see no problem of doing
that.

> Of course this is absolutely not important, it just makes life a bit
> easier. :)) By the way, the compiled driver works perfectly with JDK
> 1.4.

Strange... :) If it does not compile, it should not work. Or am I
wrong here? AFAIK, the problem is/was the incompatibility of the
signatures of couple methods in the throws clause. I wonder that
bytecode verifier does not care about such incompatibility.

Best regards,
Roman Rokytskyy