Are ZTables a free product? No. Still exists or it has just ceased to be? It is no more? It still exists, in a very domant stage. The next Zope release will probably break it right out. There's not much you can do with ZTables that you can't do with the ZCatalog.
So, ZTables is going to die silently? I don't know much about it, but considering the number of times I use TinyTable I would say that ZTables is a needed datastucture for handling larger amounts of data. Of course I understand and encourageDC to sell it as a commercial extension, but I have problems understanding why they don't release it to the public if they lost interest. Anyone from DC? regards pedro lima
Pedro Vale Lima wrote:
Are ZTables a free product? No. Still exists or it has just ceased to be? It is no more? It still exists, in a very domant stage. The next Zope release will probably break it right out. There's not much you can do with ZTables that you can't do with the ZCatalog.
So, ZTables is going to die silently?
Not necessarily.
I don't know much about it, but considering the number of times I use TinyTable I would say that ZTables is a needed datastucture for handling larger amounts of data.
Of course I understand and encourageDC to sell it as a commercial extension, but I have problems understanding why they don't release it to the public if they lost interest. Anyone from DC?
Too much effort to fix it. We don't have the time, we are working on too many projects as it is, even discussing ZTables takes more time than we have. Releasing it to the public would still take time, someone would need to maintain it, but to maintain it they'd first have to understand it, it's complex; as complex as the catalog. If we threw it out to the community I suspect it would die there also. Besides, TinyTables could be updated to support more efficient data structures like the BTrees that ZTables and ZCatalog use. ZCatalog + a more efficient TinyTables is probably 99% of ZTables. -Michel
On Thu, 30 Mar 2000, Michel Pelletier wrote:
Besides, TinyTables could be updated to support more efficient data structures like the BTrees that ZTables and ZCatalog use. ZCatalog + a more efficient TinyTables is probably 99% of ZTables.
-Michel
If one makes heavy use of TinyTables but needs more functionality (but no RDBMS) I suggest he/she takes a look at Metakit. (www.equi4.com). This amazing software deserves more attention than it gets and has already a very nice interface to python. Returned rows behave more or less like Zope Record objects. You can do sophisticated stuff like joins selects etc but it can also be used in a manner similar to TinyTable. It is transactional and *very* fast/efficient. Currently in not multithreaded so one needs to serialize requests. I think the author, Jean-Claude Wippler mentioned that he is working in a thread safe python version (the TCL one is out). A Zope interface would have beed nice ... Pavlos
On 30 Mar 2000 14:15:01 -0600, Michel Pelletier wrote:
Besides, TinyTables could be updated to support more efficient data structures like the BTrees that ZTables and ZCatalog use. ZCatalog + a more efficient TinyTables is probably 99% of ZTables.
TinyTables was originally intended to provide an easier way to manage small, static queryable tables than creating a bunch of SQL tables. It was developed for a QA-reporting system that had many static code and reference tables that would be easier to manage if they were in the object database. They turned out to be quite handy for a number of other purposes, as shown by their popularity. We considered adding more features such as record editing, etc and possibly those should be pursued. However, I suspect that TinyTables ought to remain somewhat true to its original purpose, and if more capabilities are needed, another approach should be used, even if the basic concepts remain. We wandered into the data mining realm, with excellent results on a per table basis, but since Zope doesn't really have a mechanism for connecting such objects according to the user's specification, we pretty much hit a brick wall for now. What Zope (IMO) needs is a datamining interface for all such queryable objects. Such a capability would go far beyond what both ZTables and the SQL objects now provide. We are still thinking about the issues. If you have input on this topic, I'd sure like to hear it. In this context, 'Datamining' lets the user build table object queries and see the results interactively, complete with subsetting, sorting, formatting, expression evaluation, grouping, statistical summaries, different table 'views', etc. Basically all the common stuff except table joins. This was for a population genetics research project intended to investigate the ability of users to browse and query a very large set of genetic database tables (thousands of constantly-changing tables). Somewhat similar to a data ocean as described in "Mirror Worlds" where data constantly sloshes into the ocean from the ABI labs, etc. :^)
On Tue, Apr 04, 2000 at 06:34:08PM +0000, Kent Polk wrote:
What Zope (IMO) needs is a datamining interface for all such queryable objects. Such a capability would go far beyond what both ZTables and the SQL objects now provide. We are still thinking about the issues. If you have input on this topic, I'd sure like to hear it.
You may want to look at the Gestalt System: http://gestalt-system.sourceforge.net/ -- Martijn Pieters | Software Engineer mailto:mj@digicool.com | Digital Creations http://www.digicool.com/ | Creators of Zope http://www.zope.org/ | The Open Source Web Application Server ---------------------------------------------
Martijn Pieters wrote:
You may want to look at the Gestalt System:
I browsed the info when they first announced it and looked it over again, to see if there was anything substantially different. I believe we are attempting to perform different tasks at possibly fundamentally different levels. I believe we need a more general approach to the interface problems, more like what Philip and Ty are doing with the LoginManager. I.e. make Zope smarter about how to handle its data source objects than trying to make its data source objects smarter. Of course I could be wrong in my perception of what they are trying to accomplish, but I have developed several unstructured, real-time feedback systems to investigate true Gestalt rules of perception and I don't really see a viable 'gameplan' there... Thanks Much!
participants (6)
-
Kent Polk -
kent＠tiamat.goathill.org
-
Martijn Pieters -
Michel Pelletier -
Pavlos Christoforou -
Pedro Vale Lima