preparing for new release...

NEWS.md

@@ -1,3 +1,8 @@
+# pmatch 0.1.5
+
+ * Compatibility with rlang 0.3.0
+ * Removed `cases` function (use `case_func` instead, it is much faster).
+
 # pmatch 0.1.4
 
  * New constructor code. This gives an substantial speedup when generating objects.
@@ -11,7 +16,7 @@
 # pmatch 0.1.3
 
  * Formula syntax in cases as an alternative: cases(f(x), foo ~ bar, baz ~ qux).
-   This is easier to get to static code checkes such as lintr and through the
+   This is easier to get to static code checks such as lintr and through the
    byte compiler that can complain about "assignments" to literals.
  * bind[x,y,z] <- 1:3 syntax for binding variables.
  * transformation function transform_cases_function for modifying a function

R/constructors.R

@@ -103,7 +103,7 @@ process_constructor_function <- function(constructor, data_type_name, env) {
 process_constructor_constant <- function(constructor, data_type_name, env) {
     constructor_name <- rlang::as_string(constructor)
     constructor_object <- structure(
-        NA, # FIXME: should this be list() ?
+        NA,
         constructor_constant = constructor_name,
         class = data_type_name
     )

cran-comments.md

@@ -1,35 +1,17 @@
 
-
-
 # Update to 0.1.5
 
- * New constructor code. This gives an substantial speedup when generating objects.
- * Added `case_func` as a much faster replacement for using the `cases` function. 
- * Removed `cases` (now that the faster `case_func` provides the same functionality).
- * Suggests `ggraph` and `tidygraph` for vignettes
- * Fixed a rewrite bug when using qualified names.
- * Uses `foolbox` for a safer rewrite function.
+ * Compatibility with rlang 0.3.0.
+ * Removed `cases` function (`case_func` and `case_trfunc` replace it).
 
 ## Test environments
 
 * local OS X (Mojave) install, R 3.5.1
 * ubuntu 14.04 (on travis-ci), R 3.3, 3.4, 3.5
-* win-builder (devel)
+* win-builder (devel and release)
 * RHub:
 
-  -	Windows Server 2008 R2 SP1, R-devel, 32/64 bit
-  - Windows Server 2008 R2 SP1, R-release, 32/64 bit
-  - CentOS 6, stock R from EPEL
-  - Ubuntu Linux 16.04 LTS, R-release, GCC
-  - Fedora Linux, R-devel, clang, gfortran
-  - Fedora Linux, R-devel, GCC
-  - Debian Linux, R-release, GCC
-  - macOS 10.11 El Capitan, R-release (experimental)
 
-  On Windows and Fedora, I get a Pandoc warning because 
-  it cannot find the README badges. Other than that, the 
-  checks pass.
-
 ## R CMD check results
 
 0 errors | 0 warnings | 0 notes

vignettes/getting-started-with-pmatch-.Rmd

@@ -104,7 +104,7 @@ To define a linked list type we want to specify what the basic instances look li
 linked_list := NIL | CONS(car, cdr : linked_list)
 ```
 
-Here, I have used `car` for the head of a list and `cdr` for the tail. These are traditional names inherited from the [lisp programming language that in turn inherited them from IBM 704](https://en.wikipedia.org/wiki/CAR_and_CDR). I have chosen these names so we do not class with the R functions `head` and `tail`. This expression defines the type `linked_list`, specify that we can create objects of the type using one of the two constructores `NIL` or `CONS()`, where the former takes no arguments and the second two, of which the second argument must be another linked list.
+Here, I have used `car` for the head of a list and `cdr` for the tail. These are traditional names inherited from the [lisp programming language that in turn inherited them from IBM 704](https://en.wikipedia.org/wiki/CAR_and_CDR). I have chosen these names so we do not class with the R functions `head` and `tail`. This expression defines the type `linked_list`, specify that we can create objects of the type using one of the two constructors `NIL` or `CONS()`, where the former takes no arguments and the second two, of which the second argument must be another linked list.
 
 We can create `a_list` using these constructors like this:
 
@@ -124,7 +124,7 @@ list_length <- case_func(
 list_length(a_list)
 ```
 
-When we pattern-match like this, we also bind variables. In the `CONS(car, cdr)` pattern we didn't have to name the variables `car` and `cdr` just because we called them in the constructor. In the pattern matching they are just variables and they get set to the corresponding values in the structor of `the_list`. This is why we can use `cdr` in the expression we evaluate when we match the pattern—`cdr` will be bound to the tail of the list.
+When we pattern-match like this, we also bind variables. In the `CONS(car, cdr)` pattern we didn't have to name the variables `car` and `cdr` just because we called them in the constructor. In the pattern matching they are just variables and they get set to the corresponding values in the structure of `the_list`. This is why we can use `cdr` in the expression we evaluate when we match the pattern—`cdr` will be bound to the tail of the list.
 
 We can also see this in another example where we sum all the elements in a list: