Dash for R v0.5.0#205

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

josegonzalez · 2020-05-29T00:10:19Z

Might want to add this as an enhancement :D

😹 I had forgotten to restore Percy after disabling it on a day where the tests were behaving strangely, so this was basically done in passing.

josegonzalez · 2020-05-29T00:09:51Z

What does this mean for the majority of apps? Any changes folks will need to do to make sure their apps don't do something odd?

@josegonzalez Good question; ultimately, it means that code like this:
app$layout( htmlDiv("Some text"), htmlDiv("Some more text") )
is no longer treated as valid. It now needs to be wrapped within a div, with the children passed as a list:
app$layout( htmlDiv( list( htmlDiv("Some text"), htmlDiv("Some more text") ) ) )
This was an artifact of the initial implementation (from the very, very beginning of Dash for R), which supported passing components as ... (like the R equivalent of kwargs). Realistically it's not supported by Dash, but was possible in Dash for R previously.
I'd consider this a breaking change-level modification, but now is the time to 🔪 the offending code, before we reach v1.0.

josegonzalez · 2020-05-29T00:12:05Z

Why the namespace removal?

@josegonzalez The ::: syntax is useful for accessing functions in other packages which are not exported. Usually it's something you'd apply interactively, and not a technique to apply within R package code.
What's particularly strange is that this was added to access a function within the same package, no idea why I or anyone else did this, but in any event the CRAN check is careful to disallow this behaviour for submitted packages (and with good reason).
Very much a 🙈 moment.

josegonzalez · 2020-05-29T00:12:44Z

Does the signature change impact existing calls to validate_keys that only have a single argument?

@josegonzalezvalidate_keys is an internal function, and only called from two Dash methods:
interpolate_index
index_string
The change makes it possible to use the same function for both of those methods, which pass different arguments to validate_keys (one passes the page index as a string with interpolation keys included, e.g. {%app_entry%} while the other passes in a vector of strings, in which the names of the keys are included, e.g. app_entry.
Both are matched against a list of necessary keys, and a useful error message thrown if the required keys are not present.

-Original file line number
+Diff line change
@@ Expand Up / @@ -24,7 +24,6 @@ jobs: @@
  echo "RUNNING JOB: ${CIRCLE_JOB}"
  echo "JOB PARALLELISM: ${CIRCLE_NODE_TOTAL}"
  echo "CIRCLE_REPOSITORY_URL: ${CIRCLE_REPOSITORY_URL}"
- export PERCY_ENABLE=0
  echo $CIRCLE_JOB > circlejob.txt
  git rev-parse HEAD | tr -d '\n' > commit.txt
@@ Expand Down @@

-Original file line number
+Diff line change
@@ -1,6 +1,18 @@
 # Change Log for Dash for R
 All notable changes to this project will be documented in this file.
+## [0.5.0 ] - 2020-05-28
+### Added
+- Dash for R now depends on the `brotli` package explicitly; previously it was loaded when importing `reqres`. [#204](https://github.com/plotly/dashR/pull/204)
+### Changed
+- Dash for R no longer wraps the layout in an `htmlDiv` internally, for parity with Dash for Python. Starting in v0.5.0, the `layout` method only accepts a single argument, and that argument must be a Dash component or a function that returns a Dash component. [#121](https://github.com/plotly/dashR/pull/121)
+- Package documentation has been significantly refactored to use new features of `roxygen2` when documenting R6 classes
+- The `title` method now specifies `Dash` as the default application title instead of `dash`. [#200](https://github.com/plotly/dashR/pull/200)
+### Fixed
+- A minor bug in `validate_keys` which prevented `interpolate_index` from working as intended has been resolved
 ## [0.4.1] - 2020-05-08
 ### Fixed
 - Callbacks with multiple outputs which return one or more components are now permitted; a bug previously prevented their use, fixed in [#198](https://github.com/plotly/dashR/pull/198)
@@ Expand Down @@

-Original file line number
+Diff line change
@@ -1,6 +1,6 @@
 Package: dash
 Title: An Interface to the Dash Ecosystem for Authoring Reactive Web Applications
-Version: 0.4.1
+Version: 0.5.0
 Authors@R: c(person("Chris", "Parmer", role = c("aut"), email = "chris@plotly.com"), person("Ryan Patrick", "Kyle", role = c("aut", "cre"), comment = c(ORCID = "0000-0001-5829-9867"), email = "ryan@plotly.com"), person("Carson", "Sievert", role = c("aut"), comment = c(ORCID = "0000-0002-4958-2844")), person("Hammad", "Khan", role = c("aut"), email = "hammadkhan@plotly.com"), person(family = "Plotly Technologies", role = "cph"))
 Description: A framework for building analytical web applications, Dash offers a pleasant and productive development experience. No JavaScript required.
 Depends:
@@ Expand All / @@ -20,7 +20,8 @@ Imports: @@
  digest,
  base64enc,
  mime,
- crayon
+ crayon,
+ brotli
 Suggests:
  testthat
 Collate:
@@ Expand All / @@ -38,7 +39,7 @@ License: MIT + file LICENSE @@
 Encoding: UTF-8
 LazyData: true
 KeepSource: true
-RoxygenNote: 6.1.1
+RoxygenNote: 7.1.0
 Roxygen: list(markdown = TRUE)
 URL: https://github.com/plotly/dashR
 BugReports: https://github.com/plotly/dashR/issues

-Original file line number
+Diff line change
@@ Expand Up / @@ -12,6 +12,7 @@ import(dashHtmlComponents) @@
 importFrom(R6,R6Class)
 importFrom(assertthat,assert_that)
 importFrom(base64enc,base64encode)
+importFrom(brotli,brotli_compress)
 importFrom(digest,sha1)
 importFrom(fiery,Fire)
 importFrom(fiery,combined_log_format)
@@ Expand Down @@

-Original file line number
+Diff line change
@@ Expand Up / @@ -11,4 +11,5 @@ @@
 #' @importFrom utils getFromNamespace
 #' @importFrom stats setNames
 #' @importFrom tools file_ext
+#' @importFrom brotli brotli_compress
 NULL

-Original file line number
+Diff line change
@@ Expand Up @@
 # components (TODO: this should be exported by dashRtranspile!)
 is.component <- function(x) inherits(x, "dash_component")
-# layout is really a special type of component
-is.layout <- function(x) {
- is.component(x) && identical(x[["props"]][["id"]], layout_container_id())
-}
-layout_container_id <- function() {
- "_dashR-layout-container"
-}
 # retrieve the arguments of a callback function that are dash inputs
 callback_inputs <- function(func) {
  compact(lapply(formals(func), function(x) {
@@ Expand DownExpand Up @@
  clientside_function=clientside_function
  )
  if (length(map) >= 2) {
- ids <- lapply(names(map), function(x) dash:::getIdProps(x)$ids)
- props <- lapply(names(map), function(x) dash:::getIdProps(x)$props)
+ ids <- lapply(names(map), function(x) getIdProps(x)$ids)
+ props <- lapply(names(map), function(x) getIdProps(x)$props)
  outputs_as_list <- mapply(paste, ids, props, sep=".", SIMPLIFY = FALSE)
@@ Expand DownExpand Up / @@ -1341,10 +1332,13 @@ interpolate_str <- function(index_template, ...) { @@
  return(template)
 }
-validate_keys <- function(string) {
+validate_keys <- function(string, is_template) {
  required_keys <- c("app_entry", "config", "scripts")
- keys_present <- vapply(required_keys, function(x) grepl(x, string), logical(1))
+ if (is_template)
+ keys_present <- vapply(required_keys, function(x) grepl(x, string), logical(1))
+ else
+ keys_present <- vapply(required_keys, function(x) x %in% string, logical(1))
  if (!all(keys_present)) {
  stop(sprintf("Did you forget to include %s in your index string?",
@@ Expand Down @@

Uh oh!

Uh oh!

Dash for R v0.5.0#205

Uh oh!

Dash for R v0.5.0 #205

Uh oh!

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Dash for R v0.5.0#205

Uh oh!

Dash for R v0.5.0 #205

Uh oh!

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Dash for R v0.5.0#205

Uh oh!

Dash for R v0.5.0 #205

Uh oh!

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment