v0.0.2.9023

What's Changed

• Fix file browser field - file root #358 by @DivadNojnarg in #426

Full Changelog: v0.0.2.9022...v0.0.2.9023

v0.0.2.9022

Hotfix for the add block feature in dynamic UI context (cc @JohnCoene). It was not possible to add a block to a dynamically created stack, like in the below example:

library(shiny)
library(bslib)
library(blockr)

ui <- fluidPage(
  theme = bslib::bs_theme(5L),
  actionButton("create", "Create Stack"),
  div(id = "stacks")
)

server <- function(input, output, session) {
  observeEvent(input$create, {
    stack <- new_stack()

    on.exit({
      generate_server(stack)
    })

    insertUI(
      "#stacks",
      "beforeEnd",
      generate_ui(stack)
    )
  })
}

shinyApp(ui, server)

v0.0.2.9021

What's Changed

• hotfix: registry issue (remove submit_class) by @DivadNojnarg in #422. This was preventing from seeing suggested blocks in the new add block menu after adding a submit block like filter, sumarize, ...

Full Changelog: v0.0.2.9020...v0.0.2.9021

v0.0.2.9020

Features

• Improved submit feature for blocks. Now submit isn't added as a class but as a special block attribute. When you design a block, you can pass the submit parameter like so:

new_super_block <- function(submit = NA, ...) {
  fields <- list()
  new_block(
    fields = fields,
    expr = quote(print("test")),
    submit = submit,
    ...,
    class = "my_block"
  )
}

When submit = NA, it will add a submit button but computations are blocked, as clicking on it is required. Internally, once the input$submit is clicked, the submit attribute is set to TRUE. This is useful when the stack is serialized, since this state is kept so that computations can be automatically re-triggered on restore. When submit = TRUE, a button is shown and the result is also computed. When submit = FALSE, no button is shown.

# You can disable the submit button for filter block
serve_stack(new_stack(new_dataset_block(), new_filter_block(columns = "Time", submit = FALSE)))
serve_stack(new_stack(new_dataset_block(), new_filter_block(columns = "Time", submit = NA)))
# Simulate what happens when restoring a serialised stack
serve_stack(new_stack(new_dataset_block(), new_filter_block(columns = "Time", submit = TRUE)))

v0.0.2.9010

Feature

• Improved add new block.
• Added new category to the registry. Now when a block is registered, you may pass a category parameter (which is used by the add block feature to sort blocks):

register_block(
  constructor = new_tail_block,
  name = "tail block",
  description = "return last n rows",
  category = "transform",
  classes = c("tail_block", "transform_block"),
  input = "data.frame",
  output = "data.frame"
)

If not passed, the block will belong to uncategorized blocks (default).

Doc

• Improved registry and getting started vignettes.
• Add new case studies vignette to present blockr in various contexts.
• Refine GitHub readme.

Fixes

• Fix issue in handle_remove.block: vals$stack wasn't correctly updated
  when the last block was removed leading to wrong state.
• Loading spinner is now correctly hidden when the block visual is updated.

v0.0.2

blockr 0.0.2

Breaking changes

• Change blocks and fields constructor names to new_*_block and new_*_field. For instance, for the select block,
  users are now supposed to use new_select_block() and not select_block.
• Remove data from block constructor:

new_select_block <- function(columns = character(), ...) {
  ...
}

New features

• New validation functions: validate_field() and validate_block() to check that
  values are consistent with data. These are used to then propagate any error to the user via JavaScript.
• Evaluation stops whenever a block isn't valid and the app should not crash.
• We can now instantiate block outside the stack with default parameter values (or use the old way with constructors):

# New way
data_blk <- new_dataset_block(selected = "lab", package = "blockr.data")
select_blk <- new_select_block("STUDYID")

stack <- new_stack(data_blk, select_blk)

# Old way
stack <- new_stack(new_data_block, new_select_block)

v0.0.0.9000

Initial prototype shown to the BMS leadership team. The full demo is available at: https://github.com/blockr-org/block.demo.