Hi Bob,
I'm working on fact persistence since yesterday. I've started on ATerm and XML (de)serializations of IValues.
Design decisions I've made so far:
- We allow different kinds of serialized formats via IValueReader and IValueWriter.
- IValueReader is parameterized with an IValueFactory, such that deserialization and construction is loosely coupled.
- IValues are serialized without type information (in principle). Of course, there is no rule against it, but it should not be needed.
- IValues are deserialized using an outermost Type. With this type known, and the TypeFactory
allowing queries on which other types have been defined, a deserializer can validate what it parses.
So far, this has triggered me working on gaps in our implementation of IValues. No maps, no trees, no annotations.
The reason is that I'm currently most interested in deserialization of existing XML and ATerm files, for the initial
implementation of Rascal.
On the tests, we can easily leave the tests out of the jar, but still have them in the same package. I have a good argument:
the tests being so far off, they are more likely to not be maintained.
Cheers,
Jurgen
On Thu, Oct 9, 2008 at 6:28 PM, Robert M. Fuhrer
<rfuhrer@xxxxxxxxxxxxxx> wrote:
Hi Jurgen,
Splitting out the Analysis Manager from the fact and fact base representation
code sounds fine to me.
We split the tests out into a separate project since they're not very useful to
most IMP users in a deployed environment.
BTW, is anyone over there working on fact persistence? We haven't been able
to get to it yet, but it's going to be important to us fairly soon (for search indices).
On Oct 9, 2008, at 4:18 AM, Jurgen Vinju wrote:
Hi imps,
I'd like to separate the analysis manager stuff from pdb into pdb.analysis:
- org.eclipse.imp.pdb would be for representation of, straightforward computation with, and storage of facts,
- org.eclipse.imp.pdb.analysis would be for the management of fact production and consumption
- org.eclipse.imp.pdb.ui is for visualizing facts in the IDE
- org.eclipse.imp.pdb.test is testing of org.eclipse.imp.pdb
I still would love to have the tests included with the package they belong too, please remind me why we didn't choose for that option again?
Cheers!
Jurgen
imp-dev mailing list
imp-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/imp-dev
--
Cheers,
- Bob
-------------------------------------------------
Robert M. Fuhrer
Research Staff Member
Programming Technologies Dept.
IBM T.J. Watson Research Center
IMP Project Lead (http://www.eclipse.org/imp)
X10: Productivity for High-Performance Parallel Programming (http://x10.sf.net)
_______________________________________________
imp-dev mailing list
imp-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/imp-dev
