|
|
This page will maintain a list of design issues in tZaman.
Some of the issues are OPEN, but some have been resolved and are
indicated as CLOSED.
Send e-mail to Curtis Dyreson to add an issue.
Status: OPEN |
Title: Servers passing dates to other servers |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: If a server doesn't know how to parse a date, in theory it could
hand it off to another server that does know how to parse it.
The TODO is to investigate if this is feasible, and how to set
this up in the system. |
Status: OPEN |
Title: Floating point division/multiplication |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: Currently, division like 3 days / 2 days will produce an integer
rather than a float. Add the floating point division. Ditto for
multiplication so we could do something like 1.5 * 3 days. |
Status: OPEN |
Title: Multiple Systems, unloading/loading |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: If a remote service is unavailable an exception is thrown, but
in general suppose that a server has several active services.
How do we manage the "unloading" of a calendric system that
is active in several of the services? |
Status: OPEN |
Title: CalendricSystem consistency |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: If a calendricSystem is changed on the web, when is such a change detected
by a running tauZaman server, and what is the exact semantics?
Right now it would not be detected, the server has to re-load the
system. |
Status: OPEN |
Title: URL/String passing |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: Right now the methods that pass URLs pass them as URL objects,
but we could pass them as Strings and do the URL construction
in TauZaman to be more flexible. |
Status: OPEN |
Title: Starting a service |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: The method to connect to a TauZaman server takes three parameters:
IP, name, port, but
a subset of those three should be sufficient. For instance if the port
is not given, it should default to the RMI port. Or if the IP is not given,
it should use the server name.
We should add method headers for the others. |
Status: OPEN |
Title: Schema validation |
Date: Mar-2-2006 |
Raised by: Rick |
Synopsis: We now have XML Schemas for the specification files. We should
use a validating parser to check the spec against the schema
during parsing. We need also to check how this affects
performance. |
Status: OPEN |
Title: Language property |
Date: Jan-20-2004 |
Raised by: Rick |
Synopsis: There is no language property.
I think there *should* be a language property, but this will
require redesign of some of the system, so I suggest we defer this to V2. |
Status: OPEN |
Title: Missing Epochs |
Date: Jan-20-2004 |
Raised by: Jessica |
Synopsis: We decided not to support epochs, but they appear to be useful.
How about if we add epochs? |
Status: OPEN |
Title: Allen's algebra in Period |
Date: Feb-5-2003 |
Raised by: Rick |
Synopsis: For each operation in the DeterminateSemantics/IndeterminateSemantics,
should we add a signature for that operation to the Instant, Period, or
Interval class, depending on the class of the first operand?
Currently, the operation is called through a Semantics.
We would need to add a "cached" Semantics to each constructed time value
to support this. |
Status: CLOSED |
Title: Granularity lattice |
Date: Feb-4-2003 |
Raised by: Curtis |
Synopsis: The design of the granularity lattice is open. Basically we
have to figure out how to generate and cache mappings. |
Status: OPEN |
Title: Probability Mass Function Class |
Date: Feb-4-2003 |
Raised by: Curtis |
Synopsis: Has yet to be designed. |
Status: OPEN |
Title: System Testing |
Date: Feb-4-2003 |
Raised by: Curtis |
Synopsis: We have to develop a comprehensive set of tests. There should
be both stress tests and performance tests. TimeADT has some test code.
Initially, the code specified there should be adapted. |
Status: OPEN |
Title: Full elaboration of Semantics |
Date: Feb-4-2003 |
Raised by: Curtis |
Synopsis: The IndeterminateSemantics and DeterminateSemantics lack an
exhaustive set of operations, for instance there is no overlaps
constructor (there is a predicate, but no constructor).
All operations need to be fully elaborated. |
Status: CLOSED |
Title: Logo Design |
Date: Jan-31-2003 |
Raised by: Curtis |
Synopsis: The logo needs to be redesigned to make the dates legible. |
Status: OPEN |
Title: Use Cases |
Date: Jan-31-2003 |
Raised by: Curtis |
Synopsis: We need to develop a use case document. Jessica added that
the tool she uses for the UML diagrams can generate some use cases.
We should definitely incorporate as many automatically generated
cases as possible. But we also should hand-develop some use cases.
The idea is a set of examples that a user, interested in particular
components, can use to further write his/her own code. Bedirhan
is to develop use cases for the Server/CalendricSystem. Curtis is
to develop use cases for timestamp/temporaldatatypes. |
|
|