* don't assume all .so files are the exectuable
* don't rebuild support or the compiler if only the buildIdris nix function has changed
* fixes to nix buildIdris function
* [ cleanup ] Move repeating trying-with-unambiguity pattern to a function
* [ fix#2850 ] Search for dependencies for default hints among all hints
* [ fix#2932 ] Always try defaults in dependencies search
When implicitly binding a variable, use the location of the head
function/constructor that expects it. This way we can differentiate
multiple implicits bound on the same LHS.
Note that this does not resolve the issue 34: there the location
is then further muddled by the fact that where-bound functions
are lifted to the toplevel.
* [ fix#72 ] remove the broken modules
People are still hitting the same issueT
There has been no movement towards fixing it
It is IMO unfixable
Let's drop it.
* [ fix#72 ] Remove dependencies of Control.Algebra
Follow-up to the commit by gallais, this removes the contrib libraries
which were using `Control.Algebra`.
* [ fix#72 ] Record changes in CHANGELOG_NEXT
* [ lint ] Move Algebra changes to existing header
---------
Co-authored-by: Thomas E. Hansen <teh6@st-andrews.ac.uk>
* don't show 'Just' before package versions
* unrelated - fix nix develop shell for darwin
* spit out the package path for each package as well as the package name and version
* put the package path on a new line
* add TTC version info to output
* don't name unused variable
* add changelog entry
* Initial stab at package json dump
* expose ipkg json dump as new option
* make dependency output easier to ingest by another tool
* Add a test for ipkg json dump
* cleanup
* maybe just don't collide with existing equally good fixity
* make new operator private
* Add new module to api ipkg file
* Add note to CHANGELOG_NEXT
* correct the docs for the dump-ipkg-json command
This makes it possible to call the function in more situations. It also
brings its signature in line with the overloads on `List`, `List1` and
`SnocList`.
The previous implementation of `Data.Vect.Elem.dropElem` required the
length of the `Vect` to be available at runtime. This was used in order
to recurse in the case that the `Elem` is not `Here`. However, it turns
out that this is not actually necessary. Idris can deduce that the tail
must be non-empty if it contains an `Elem`.
* [ refactor ] moving Data.OpenUnion to base
* [ new ] System.Concurrency.(Linear/Session)
* [ test ] for the new feature
Fixing other tests impacted by the refactoring
* [ cleanup ] move definitions around, touch up docs
* [ fix ] re-export linear notations for Control.Linear.LIO
The previous command does not work in any shell I am aware of.
The new one directly opens the library documentation in $BROWSER if it
is set and otherwise just echos the path to index.html.
* [ base ] Data.SortedSet.leftMost and .rightMost
Implement `leftMost` and `rightMost` for `SortedSet` in terms of the
functions with the same name in `Data.SortedMap`.
* contributors
---------
Co-authored-by: G. Allais <guillaume.allais@ens-lyon.org>
* [ performance ] a much faster implementation of unpack
* [ fix ] partiality error in Hangman test
* [ test ] add some documentation to the unpack test case
* [ test ] fix expected output of unpack test
