| Author |
Message |
|
|
Can you print other web pages in colour using the same computer, same web browser, and same printer?
-Jonathan
|
 |
|
|
Point taken.
We have started talking more seriously about switching our codebase to Java 6 in the near term. The catalyst was the release of OS X Snow Leopard, which (finally!) supports Java 6 on all Intel Mac hardware, including 32-bit. It also makes OS X Tiger (which is artificially stuck on Java 5 by Apple) 2 generations old.
Why do we care about Mac compatibility so much? Take a look at the download numbers for 0.9.13:
Windows: 9610
Mac: 3090
Others: 3165
Compared to their overall market share, Mac and Linux/BSD/Solaris are way overrepresented. I think it's wise for us to respect that.
Of course, we can't tell how many of the Mac downloads were for 10.4, 10.5, and 10.6. We can only hope the 10.4 systems have mostly been replaced by now.
-Jonathan
|
 |
|
|
Due to licensing restrictions and file size, SQL Power's open source tools don't ship with a copy of the Oracle JDBC driver. That means you have to provide your own copy of the Oracle driver and tell Wabit where to find it. See Chapter 4 of the Wabit user guide (available under the Help menu) for details on how to do this. You'll find specific step-by-step instructions in the subsection, "Defining the JDBC Driver."
-Jonathan
|
 |
|
|
Thank you for taking the time to report this problem. I've filed a bug detailing the issue.
-Jonathan
|
 |
|
|
- Tables that don't exist should be added automatically (or the user should be asked if he wants to add them)
It actually had this feature at first, but I took it away because I realized it was a bit redundant given that you're inside a data modeling tool with pretty decent reverse/forward engineering capabilities. Just create a new project, drag in the tables you want, adjust things you don't like, and forward engineer to the target.
Or, if the target is already similar to the source, use the Compare DM tool to synchronize the target structure.
These are things that would have to be reinvented if they were embedded in the Copy Table Data feature. Especially the ability to tweak before forward engineering is quite handy.
Of course, you might disagree, and you might be right! Feel free to make a case for your suggestion.
- Rename Cancel button to Close (I don't want to cancel I just want to close after I'm done with copying).
That would be better! Thank you.
- Copying multiple tables at the same time didn't work for me.
It works, but the UI is non-obvious. I think the user guide explains the correct way of doing this. It's deceptively powerful: when copying multiple tables at a time, the Architect actually analyzes the FK dependencies and automatically loads parent tables before their children.
The tool helped me a lot after I figured how to use it!
Glad to hear it. If you can suggest how to make the intent of the UI more obvious now that you figured out how it works, please let us know. We're all for improving it. Just between you, me, and the Google bot, I actually slapped this feature together for my own purposes a few years back, intending to hide it from the GUI before releasing to the public (pending a proper UI design), then I forgot to hide it. Now, years later, it remains unhidden but still undesigned too.
-Jonathan
|
 |
|
|
Hi Kevin,
Thanks for your question. It's one of the most frequently requested features, and we do have a plan in place to give you access to all native data types without compromising one of the Architect's distinguishing features: letting your keep your data model platform-independent.
See my response to this similar question for a more complete response: http://www.sqlpower.ca/forum/posts/list/2060.page
When the domains feature is implemented, it will be your choice whether or not you want your data model to be platform neutral. You'll be able to create a domain/user-defined type called "NVARCHAR(x)" if you want to, but you'd also be able to create more meaningful logical domains such as "Street Address" and then direct Architect to forward-engineer that type in SQL Server 2008 as an NVARCHAR(200). Your choice, but either way you get access to all the native types you need.
-Jonathan
|
 |
|
|
Sounds like a good idea. I've filed an enhancement request that covers everything you've asked for, but is generalized so it will work for all platforms rather than being Oracle-specific.
Thanks for taking the time to post!
-Jonathan
|
 |
|
|
Hi there,
We don't have any class diagrams of Power*Architect's implementation. I guess that's kind of surprising, considering the fact that it's a tool for making the same types of diagrams for database schemas.
If you're planning to contribute back the changes you're making, we'd be happy to assist you by answering any questions you have about the Architect's internal API. In order to accept your changes, we'd ask you to sign our contributor agreement, which is based on the Sun SCA: you retain full rights to the code you contribute, and we also receive full rights to it. And of course we will weigh your contribution on its merits. We look for correct use of our code formatting style (documented on the Wiki), following of established idioms, user interface consistency with the rest of the app, fit with current and future plans, and so on. This feature definitely fits with our plans, so we're interested!
To get started with visual aliases, you'll want to understand the SQLTable class and the TablePane class. The former is the non-GUI data model of a table in Architect, and the latter is the GUI component that lives in the PlayPen. It is perfectly permissible to attach multiple TablePane instances to a single SQLTable.
Similarly, there is a SQLRelationship class that models an FK constraint, and a Relationship class that implements the visual component. Again, there is no issue attaching multiple Relationship components to a single SQLRelationship.
So that part is easy. The duplicate GUI components will even save and load properly. The trick will be to do a good UI around this concept so it doesn't become unnecessarily confusing when there are clones/aliases of tables floating around the playpen.
Thanks for your interest. If you'd like to contribute your changes, just let us know and we'll send you a contributor agreement form.
-Jonathan
|
 |
|
|
Ok, glad to hear you got it working.
My experiences with community failure were with JFreeReport, the overall Pentaho reporting platform, and the Mondrian users forum. I haven't used Kettle enough to patch bugs or require community support, but I'm glad to hear that the Kettle community (and Matt) is responsive.
About Mondrian, although we get the silent treatment when asking direct Mondrian questions on the Mondrian forums, I can vouch for Julian Hyde as being very conscientious and helpful when left to his own devices: he's always willing to help on they users and developers list of his olap4j project at sourceforge. But that project is not Pentaho-sponsored.
Thanks for your interest in Wabit! Please let us know if you run into any more bugs or misfeatures.
-Jonathan
|
 |
|
|
Hi Tom, It looks as if your format string is being interpreted as two space-separated format strings: # and ###,###. I'm guessing your format string is meant to cause the number 123456789.12345 to come out as 123 456 789,123 but you'll actually see 123456789 123,456,789. Wabit actually relies on the olap4j driver to perform the formatting of values in OLAP result sets. Perhaps this is a bug in the Mondrian olap4j driver, or in Mondrian itself. One workaround I could suggest would be to use a character other than space in your format string. Non-breaking space (Unicode 00A0) seems a good candidate: formatString="# ##0,###" [edit: escaped the ampersand--apparently JForum doesn't escape XML entity references] You may also want to file a bug or ask about this on the Mondrian-users mailing list. Be warned, however, I have never received a reply to any question, bug report, or patch I have submitted to Pentaho as a "community" member. I've only been able to solicit responses from them within the context of being a business partner. -Jonathan
|
 |
|
|
We could also use a double mouse click on a table to expand it and to minimize it.
That's my favourite option so far. It's an expert feature. I'm not convinced advertising it visually would achieve anything but to add visual noise to the GUI. (but once a table is rolled up/collapsed/condensed, it is then important to indicate that state visually)
I don't know if I'll ever have that much time to program, but i would love to.
As the charities are so fond of saying, "every little bit helps." You don't have to tackle the big items, but it would be a huge help if you could implement some of the smaller changes you've suggested. You're already a project member, so no need for sending patches. Just commit. If we dislike something or find a problem, we'll be sure to let you know.
-Jonathan
|
 |
|
|
Great. Here are some pointers:
Instructions on how to check out and build the Architect codebase: http://code.google.com/p/power-architect/wiki/GettingStarted
Then take a look at the existing DB2 DDL generator and the HSQLDB DDL generator. These are pretty simple ones. SQLite will probably be similar. For a more involved generator, look at the PostgreSQL and/or SQL Server ones.
Be careful about calls to SQLObject.getName() vs. calls to SQLObject.getPhysicalName(). There are still some bugs lurking in there due to an incomplete change to the way we handle physical names. The (new) correct approach is to use getPhysicalName() for everything, and not to modify the physical name (some old DDL generation code still calls setPhysicalName(), which is now wrong).
-Jonathan
|
 |
|
|
Perhaps give a platform-specific generator a try. The generic one tries to determine type names and other features from DatabaseMetaData, which is probably returning null from some calls that aren't supposed to return null. The DB2 and HSQLDB ones are pretty similar to the generic one, but they don't ask the driver any questions.
We'd gladly accept a SQLiteDDLGenerator class if you're so inclined to produce one*. You could use the nearest existing generator implementation as a starting point.
*as long as you're willing to sign a SQL Power contributor agreement
-Jonathan
|
 |
|
|
Hi Andreas,
Caused by: java.net.ConnectException: Connection refused
This can only mean that the remote server is refusing the TCP connection request. Go over your SQL Server (and firewall) configurations again; there must be something missing.
-Jonathan
|
 |
|
|
Cool, thanks. I found your message. Although I don't see the patch file itself in the list archive.
I noticed there were only 4 posts on that SQLiteJDBC list for the month of August, and no responses at all. So I wouldn't take the "silent treatment" too personally.
About the null pointer exception upon forward engineering, did you choose a default DDL generator when setting up the SQLite database type? That may be the cause of the problem. If so, the error message in Architect can/should be a whole lot better.
-Jonathan
|
 |
|
|