ddlog

ddlog copied to clipboard

Compiler ignores constants in head atoms of an inference rule

2

This rule ``` @weight(f) tag(w, "GROOVY") :- features(w, f). ``` is compiled as if it's ``` @weight(f) tag(w, t) :- features(w, f). ```

alldefector

(At least in the dynamic multinomial branch) the spouse example can barely finish because the grounding SQL of a binary factor is on top of a double-self-join where each base...

alldefector

Currently SQL functions such as random() that has zero arguments are not accepted using deepdive compile. Here is Jaeho's note: > Actually, I took a brief look at ddlog compiler...

xiao-cheng

include/import other files

6

To be more pragmatic for real users, sharing common schema declaration, etc. will be necessary. One easy way to implement this is with the `cpp` C/C++ macro processor and let...

netj

I have a ddlog file shown below. Note that the last reference to the nlp table has the wrong schema. When the ddlog file is compiled, there was no error...

xiaoling

Cannot combine array indexing with minus

3

Just tried writing a query like `blah(wordidxs[1] - 3) :- ....` , but cannot combine `-` with array indexing

Colossus

The types used in function (udf) declaration should accept two forms: - [x] 1. Using `rows like X` (similar to the currently only supported `like X` syntax): ``` function f...

netj

If table defined first as view, then table, view isn't dropped

1

In ddlog, if you don't specify a table layout, arbitrary queries are being set up as views. If you later decide to "materialize" the view by writing a table definition...

Colossus

Some important warnings should fail the build, such as non-exaustive pattern matching, motivated by #77.

netj

DDlog parser needs a rewrite using a faster parser combinator library

10

I first thought it was a JVM booting latency issue and optimized that a bit, but didn't help much (1-2 seconds saving after first launch). I just did a dumb...

netj