From: Don A. <dal...@us...> - 2002-05-19 15:13:42
|
The ZODB implementation is slowly coming along. Memory usage is typically significantly lower than the XML implementation, and doesn't seem to get much worse as the database size increases. Load times are much quicker, typically taking less than 10-15 seconds for very large databases, as opposed to 1-2 minutes. File size is significantly larger. While the XML database is usually less than 200KB, the ZODB database is usually in multiple megabyte range. Also, the XML database can be examined and edited by hand. The ZODB implementation is still not ready for primetime, but it is improving. I still believe that for the majority of users (those with databases under about 5000 people) that the XML database will be the best. But for large databases, the ZODB interface will be a necessity. Don |
From: Shane H. <sh...@zo...> - 2002-05-20 15:12:01
|
Don Allingham wrote: > The ZODB implementation is slowly coming along. Memory usage is > typically significantly lower than the XML implementation, and doesn't > seem to get much worse as the database size increases. Load times are > much quicker, typically taking less than 10-15 seconds for very large > databases, as opposed to 1-2 minutes. > > File size is significantly larger. While the XML database is usually > less than 200KB, the ZODB database is usually in multiple megabyte > range. Also, the XML database can be examined and edited by hand. In my limited testing, using PersistentList (as opposed to simple lists) caused the database size to expand dramatically. This is understandable since without PersistentList, each person object takes one database record, but with PersistentList, each person object takes several records. Have you tried it without PersistentList? > The ZODB implementation is still not ready for primetime, but it is > improving. > > I still believe that for the majority of users (those with databases > under about 5000 people) that the XML database will be the best. But for > large databases, the ZODB interface will be a necessity. Also note that with ZODB you will be able to add an "undo" feature. Shane |
From: Don A. <dal...@us...> - 2002-05-21 00:35:11
|
On Mon, 2002-05-20 at 09:11, Shane Hathaway wrote: > In my limited testing, using PersistentList (as opposed to simple lists) > caused the database size to expand dramatically. This is understandable > since without PersistentList, each person object takes one database > record, but with PersistentList, each person object takes several > records. Have you tried it without PersistentList? I've been using PersistentList because some of the documentation that I read indicated that if I use a normal list, ZODB would not detect a change. So I've been using the PersistentList until I get a better understanding of what I am doing. Unfortunately, I haven't found too much in the way of documentation yet. Don -- Don Allingham dal...@us... http://gramps.sourceforge.net |
From: Shane H. <sh...@zo...> - 2002-05-21 00:47:30
|
On 20 May 2002, Don Allingham wrote: > On Mon, 2002-05-20 at 09:11, Shane Hathaway wrote: > > In my limited testing, using PersistentList (as opposed to simple lists) > > caused the database size to expand dramatically. This is understandable > > since without PersistentList, each person object takes one database > > record, but with PersistentList, each person object takes several > > records. Have you tried it without PersistentList? > > I've been using PersistentList because some of the documentation that I > read indicated that if I use a normal list, ZODB would not detect a > change. So I've been using the PersistentList until I get a better > understanding of what I am doing. Unfortunately, I haven't found too > much in the way of documentation yet. There's some great documentation here: http://www.zope.org/Documentation/ZDG/Persistence.stx I added the "self._p_changed = 1" everywhere a list attribute is modified. That causes ZODB to see the change. The only problem is that the actual lists are returned from some accessor methods, and that's not good since application code might change the lists without notifying ZODB. That's why you might consider returning tuples or copies of the lists instead. Shane |
From: Mark L. <ma...@fo...> - 2002-05-25 19:18:37
|
Hello, I've been looking through the Gramps source from CVS over the last week or two, and I wonder if I can be of any help in the development. By way of introduction, I'm a programmer out in Chicago. I usually program in Java and C++ for work, but I've been doing some Python development at home-- I wrote a raytracer a couple of months ago as a learning exercise, some simple GNOME programs using libglade, that sort of thing. I was wondering, what are the current design goals? From the recent posts to the mailing list and from the CVS commits, it looks like most of the recent development is going into ZODB. I'm not familiar with ZODB, so was wondering if there is something else I could start looking at. Thanks for your time! -- Mark Lewis ma...@fo... |
From: Don A. <dal...@us...> - 2002-05-25 19:50:41
|
Mark, Help is always appreciated. There are lots of available things that need to be done. These include: 1) Improve the FamilyView, so that few mouse clicks are needed. 2) Be able to export a database to an ISO-9660 image that can be burned onto a CDROM. 3) Implement a GNTP client (a peer-to-peer genealogy protocol) 4) Look at a GNOME 2 port 5) Improve reports 6) Be able to translate a set of locations onto a graphical map (so you can see where in the world certain people came from). 7) Enhancing the relationship calculator to handle cultural/language differences. 8) Enhance the Date/Calendar structure so that alternate calendars can be plugged into the code, instead of the 4 hardcoded calendars available now (Julian, Gregorian, Hebrew, and French Republican) 9) Provide a python interface to the graphviz layout algorithms (www.graphviz.org) or provide some other graph layout mechanism. 10) Anything else that matches your interest. Welcome aboard. Don On Sat, 2002-05-25 at 13:19, Mark Lewis wrote: > Hello, > > I've been looking through the Gramps source from CVS over the last week > or two, and I wonder if I can be of any help in the development. > > By way of introduction, I'm a programmer out in Chicago. I usually > program in Java and C++ for work, but I've been doing some Python > development at home-- I wrote a raytracer a couple of months ago as a > learning exercise, some simple GNOME programs using libglade, that sort > of thing. > > I was wondering, what are the current design goals? From the recent > posts to the mailing list and from the CVS commits, it looks like most > of the recent development is going into ZODB. I'm not familiar with > ZODB, so was wondering if there is something else I could start looking > at. > > Thanks for your time! > > -- Mark Lewis > ma...@fo... > > > _______________________________________________________________ > > Don't miss the 2002 Sprint PCS Application Developer's Conference > August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm > > _______________________________________________ > Gramps-devel mailing list > Gra...@li... > https://lists.sourceforge.net/lists/listinfo/gramps-devel -- Don Allingham dal...@us... http://gramps.sourceforge.net |