graceful-project / grace Goto Github PK

Currently our solver outputs a result with the following format:
'{"result":"[{\"rainfall4\" : 10.0,\n\"inflow5\" : 10.0,\n\"outflow5\" : 10.0,\n\"inflow6\" : 10.0}\n]"}'
where "result" is a string instead of a proper JSON object.

We should be able to change this with some simple modifications to the Res type and submit function in RestAPI/Main.hs.

We should be able to define new libraries by combining and extending previously defined libraries (without having to copy all the code over or alter the original library file).

Use hint to interpret library components written by the user. Here is an idea for how to do it: https://github.com/MaximilianAlgehed/TestStackThing

Hello,

There is potentially going to be more than 5 people working on this code base.
To ensure that this works out well Björn and I discussed setting up some guidelines
and routine to make sure that new contributions do not break the library etc.

Maybe we need to have a short meeting to make sure everyone agrees on what
the rules are etc. I would like to do this before the holiday season.

Any comments and thoughts would be appreciated.

To reproduce:

stack build
stack exec Examples

Output:

Results via old compiler:
{"total cost" : 27,
"pump operation" : 3,
"pump increased" : 1,
"storage increased" : 13,
"overflow" : 0,
"inflow" : 20,
"is flooded" : false}
,

Results via haskelzinc:
model.mzn:25:
constraint 0 < v11 = v12;
                   ^
Error: syntax error, unexpected =

Examples: callCommand: mzn2fzn model.mzn (exit 1): failed

Can we fix the module structure, something like:

GCM.hs
CP.hs
Compiler.hs
Compiler/
    Compiler0.hs
GRACe.hs
GCMP.hs
GCMP/
    GCMPInternal.hs
    GCMPTesting.hs

eugraceful/generic-library := eugraceful/grace-examples

Also please share the scripting needed (dockerfile etc.) for future uploads.

Please use the text from the deliverable to update the install instructions in the README file to start with the easy way (using docker) and only then describe "developer install instructions" (~= what is there now).

/P

We need to include the CLD code as a library and document with the VisualEditor group how to translate a CLD to these components.

Hello everyone,

I've been trying to get MiniZinc working on a new computer
and there is no_way_in_hell to get it to work. All distributions,
from the one in the debian repositories to the version on the
MiniZinc website, are broken.

Has anyone managed to compile MiniZinc from source?
Does anyone know how to get everything to work?

• Our README.md is not up to date, it has intructions that no longer work and lacks information about the web service.
• Need to go through code and update documentation.

Parameters not attached to actions behave like
unconstrained decision variables as a result of
the "a#" variable not being set to false. Someone
fix please...

GenericLibrary := GRACe

I was denied permission to push to this repo, can somebody look into/fix that? Adding it to the DSL team might be a good idea.

We are setting out to write an optimising compiler. If we are going to do this correctly we will need
a test suite to make sure our optimisations are correct. I suggest some form of property based testing (obviously). One way we might be able to do this in an innovative way is to create a bunch of GCM components and specify their behaviour in terms of functions from "bound" (or restricted) constraint variables to "unbound" ones. Using this we can check that the solutions generated by MiniZinc are valid
for our examples.

While this may provide property based testing in some sense, maybe we should also test each new iteration of the compiler against some old (this current) version for a bunch of random programs, to make sure that the semantics don't change with each iteration.

Tell me what you think.

Make sure they're consistent with what is described here:
https://github.com/GRACeFUL-project/DSL-internal/blob/master/ref/qpn.pdf
https://github.com/GRACeFUL-project/DSL-internal/blob/master/ref/QPNModelerTests.pdf
and give the expected outcome on examples from van Kouwen's thesis.

To avoid confusion like we had last night!