New Features

  • Now plyr (>= 1.8.7) supports progressr for also parallel processing, e.g. y <- plyr::llply(X, slow_sum, .parallel = TRUE, .progress = "progressr").

Bug Fixes

  • The ‘plyr’ progress plugin stopped working with progressr 0.8.0.

  • Warnings on stray progression conditions could appear with an empty message.

Significant Changes

  • Now interrupts are detected, which triggers the progress handlers to terminate nicely, e.g. a progress bar in the terminal will stay as-is instead of being cleared.

Bug Fixes

  • A progressor that signaled progress beyond 100% prevented any further progressors in the same environment to report on progress.

  • It was not possible to reuse handlers of type ‘progress’ more than once, because they did not fully reset themselves when finished.

  • The ‘pbcol’ progression handler did not respect clean = FALSE.

Deprecated and Defunct

  • Function progress() is defunct in order to re-use it for other purpose. It is unlikely that anyone really used this function, but if you did, then use cond <- progression() to create a progression condition and then use withRestart(signalCondition(cond), muffleProgression = function(p) NULL) to signal it.


  • The progressor function created by progressor() no longer “inherit” objects from the calling environment, which would, for instance, result in those objects to be exported to parallel workers together with the progressor function, which in turn would come with large time and memory costs.

  • progressor() no longer records the call stack for progressions by default, because that significantly increases the size of these condition objects, e.g. instead of being 5 kB it may be 500 kB. If a large number of progress updates are signaled and collected, as done, for instance, by futures, then the memory consumption on the collecting end could become very large. The large sizes would also have a negative impact on the performance in parallelization with futures because of the extra overhead of transferring these extra large conditions from the parallel workers back to the main R session. These issues has been there since progressr 0.7.0 (December 2020). To revert to the previous behavior, use progressor(..., trace = TRUE).

New Features

  • progressor() gained argument trace to control whether or not the call stack should be recorded in each progression condition.

  • Now print() for progressor functions and progression conditions report also on the size of the object, i.e. the number of bytes it requires when serialized, for instance, to and from a parallel worker.

Bug Fixes

  • Registered progression handlers would report on progress also when in a forked parallel child processes, e.g. when using parallel::mclapply(). This would give a false impression that progressr updates would work when using parallel::mclapply(), which is not true. Note however, that it does indeed work when using the future ‘multicore’ backend, which uses forks.

Significant Changes

  • Creating a new progressor() will now automatically finish an existing progressor, if one was previously created in the same environment. The previous behavior was to give an error (see below bug fix).

  • R_PROGRESSR_* environment variables are now only read when the progressr package is loaded, where they set the corresponding progressr.* option. Previously, some of these environment variables were queried by different functions as a fallback to when an option was not set. By only parsing them when the package is loaded, it decrease the overhead in functions, and it clarifies that options can be changed at runtime whereas environment variables should only be set at startup.

  • When using withProgressShiny(), progression messages now updates the detail component of the Shiny progress panel. Previously, it updated the message component. This can be configured via new inputs argument.

New Features

  • withProgressShiny() gained argument inputs, which can be used to control whether or not Shiny progress components message and detail should be updated based on the progression message, e.g. inputs = list(message = "sticky_message", detail = "message") will cause progression messages to update the detail component and sticky ones to update both.

  • Now supporting zero-length progressors, e.g. p <- progressor(along = x) where length(x) == 0.

  • Add handlers("rstudio") to report on progress in the RStudio Console via the RStudio Job interface.

Beta Features

  • As an alternative to specifying the relative amount of progress, say, p(amount = 2), it is now possible to also specify the absolute amount of progress made this far, e.g. p(step = 42). Argument amount has not effect when argument step is specified. WARNING: Argument step should only be used when in full control of the order when this progression condition is signaled. For example, it must not be signaled as one of many parallel progress updates signaled concurrently, because we cannot guarantee the order these progressions arrive.

Bug Fixes

  • In progressr 0.7.0, any attempt to use more than one progressor inside a function or a local() call would result in: “Error in assign(”…progressor”, value = fcn, envir = envir) : cannot change value of locked binding for …progressor.”

Deprecated and Defunct

  • Function progress() is deprecated in order to re-use it for other purpose. It is unlikely that anyone really used this function, but if you did, then use cond <- progression() to create a progression condition and then use withRestart(signalCondition(cond), muffleProgression = function(p) NULL) to signal it.

Significant Changes

  • The user can now use handlers(global = TRUE) to enable progress reports everywhere without having to use with_progress(). This only works in R (>= 4.0.0) because it requires global calling handlers.

  • with_progress() now reports on progress from multiple consecutive progressors, e.g. with_progress({ a <- slow_sum(1:3); b <- slow_sum(1:3) }).

  • A progressor must not be created in the global environment unless wrapped in with_progress() or without_progress() call. Ideally, a progressor is created within a function or a local() environment.

  • Package now requires R (>= 3.5.0) in order to protect against interrupts.

New Features

  • progressor() gained argument enable to control whether or not the progressor signals progression conditions. It defaults to option progressr.enable so that progress updates can be disabled globally. The enable argument makes it easy for package developers who already provide a progress = TRUE/FALSE argument in their functions to migrate to the progressr package without having to change their existing API, e.g. the setup becomes p <- progressor(along = x, enabled = progress). The p() function created by p <- progressor(..., enable = FALSE) is an empty function with near-zero overhead.

  • Now with_progress() and without_progress() returns the value of the evaluated expression.

  • The progression message can now be created dynamically based on the information in the progression condition. Specifically, if message is a function, then that function will called with the progression condition as the first argument. This function should return a character string. Importantly, it is only when the progression handler receives the progression update and calls conditionMessage(p) on it that this function is called.

  • progressor() gained argument message to set the default message of all progression updates, unless otherwise specified.

  • progressor() gained argument on_exit = TRUE.

  • Now the progress handler shows also a spinner by default.

  • Add the ‘pbcol’ handler, which renders the progress as a colored progress bar in the terminal with any messages written in the front.

  • Progression handlers now return invisibly whether or not they are finished.

Bug Fixes

  • Zero-amount progress updates never reached the progress handlers.

  • Argument enable for with_progress() had no effect.

Significant Changes

  • Now with_progress() makes sure that any output produced while reporting on progress will not interfere with the progress output and vice versa, which otherwise is a common problem with progress frameworks that output to the terminal, e.g. progress-bar output is interweaved with printed objects. In contrast, when using progressr we can use message() and print() as usual regardless of progress being reported or not.

New Features

  • Signaling progress(msg, class = "sticky") will cause the message to be sticky, e.g. for progress bars outputting to the terminal, the message will be “pushed” above the progress bar.

  • with_progress() gained argument delay_terminal whose default will be automatically inferred from inspecting the currently set handlers and whether they output to the terminal or not.

  • Arguments delay_stdout and delay_conditions for with_progress() is now agile to the effective value of the delay_terminal argument.

  • Now handler_nnn() functions pass additional arguments in ... to the underlying progress-handler backend, e.g. handler_progress(width = 40L) will set up progress::progress_bar$new(width = 40L).

  • Add environment variables R_PROGRESSR_CLEAR, R_PROGRESSR_ENABLE, R_PROGRESSR_ENABLE_AFTER, R_PROGRESSR_TIMES, and R_PROGRESSR_INTERVAL for controlling the default value of the corresponding progressr.* options.

Bug Fixes

  • Limiting the frequency of progress reporting via handler arguments times, interval or intrusiveness did not work and was effectively ignored.

  • The progress handler, which uses progress::progress_bar(), did not support colorization of the format string when done by the crayon package.

  • handlers() did not return invisible (as documented).

  • Argument target was ignored for all handler functions.

  • Argument interval was ignored for handler_debug().

  • The class of handler_<nnn>() functions where all reset_progression_handler rather than <nnn>_progression_handler. The same bug caused the reported name field to be "reset" rather than "<nnn>".

New Features

  • Add ‘void’ progression handler.

Bug Fixes

  • Only the last of multiple progression handlers registered was used.

Significant Changes

  • All progression handler function have been renamed from <name>_handler() to handler_<name>() to make it easier to use autocompletion on them.

New Features

  • progressor() gained arguments offset and scale, and transform.

  • handlers() gained argument append to make it easier to append handlers.

Bug Fixes

  • A progression condition with amount = 0 would not update the message.

Bug Fixes

  • winprogressbar_handler() would produce error “invalid ‘Label’ argument”.

  • handlers() did not return a list if the ‘default’ handler was returned.

Significant Changes

New Features

  • handlers() gained argument default specifying a progression handler to be returned if none is set.

New Features

  • Add withProgress2(), which is a plug-in backward compatibility replacement for shiny::withProgress() wrapped in progressr::with_progress() where the the “shiny” progression handler is by default added to the list of progression handlers used.

  • Add demo("mandelbrot", package = "progressr").

Bug Fixes

  • Package could set .Random.seed to NULL, instead of removing it, which in turn would produce a warning on “‘.Random.seed’ is not an integer vector but of type ‘NULL’, so ignored” when the next random number generated.

New Features

  • Add support for progressor(along = ...).

New Features

  • Now it is possible to send “I’m still here” progression updates by setting the progress step to zero, e.g. progress(amount = 0). This type of information can for instance be used to updated a progress bar spinner.

  • Add utility function handlers() for controlling option progressr.handlers.

  • Progression handlers’ internal state now has a sticky message field, which hold the most recent, non-empty progression message received.

New Features

  • with_progress() gained arguments enable and interval as an alternative to setting corresponding options progressr.*.

  • Now option progressr.interval defaults to 0.0 (was 0.5 seconds).

  • Added print() for progression_handler objects.

Bug Fixes

  • with_progress(..., delay_conditions = "condition"), introduced in progressr 0.1.0, would also capture conditions produced by progression handlers, e.g. progress::progress_bar() output would not be displayed until the very end.

New Features

  • with_progress() now captures standard output and conditions and relay them at then end. This is done in order to avoid interweaving such output with the output produced by the progression handlers. This behavior can be controlled by arguments delay_stdout and delay_condition.

New Features

  • Now a progression condition is identified from the R session UUID, the progressor UUID, the incremental progression index, and the progression timestamp.

Bug Fixes

  • A progressor object that was exported to the same external R process multiple times would produce progression conditions that was non-distinguishable from those previously exported. Adding a timestamp to the progression condition makes them distinguishable.

New Features

  • Add print() for progression conditions and progressor functions.

  • Now the progressors record more details on the session information. This information is passed along with all progression conditions as part of the internal owner information.

New Features

  • Add filesize_handler progression handler.

  • Add support for times = 1L for progression handlers which when used will cause the progression to only be presented upon completion (= last step).

  • The shutdown control_progression signaled by with_progress() on exit now contains the status of the evaluation. If the evaluation was successful, then status = "ok", otherwise "incomplete". Examples of incomplete evaluations are errors and interrupts.

New Features

Software Quality

  • TESTS: Increased package test coverage of progression handlers by running all code except the last step that calls the backend, which may not be installed or supported on the current platform, e.g. tcltk, beepr, notifier.

Bug Fixes

  • Precreated progression handlers could only be used once.

  • with_progress(..., cleanup = TRUE) requires a withRestart() such that also “shutdown” progressions can be muffled.

New Features

  • Add argument enable_after for progression handlers.

  • Now with_progress(..., cleanup = TRUE) will signal a generic “shutdown” progression at the end that will trigger all progression handlers to finish up regardless of all steps have been take or not.

  • Now progressions originating from an unknown source are ignored.

  • The default output format of the progress::progress_bar() progression handler is now ":percent :bar :message".

  • The tcltk::tkProgressBar() progression handler now displays the progression message.

  • Now the progression condition itself is passed to the progression reporter functions.

  • Add ‘debug_handler’ for prototyping and debugging purposes.

  • Add ‘newline_handler’ to add newlines between output of multiple handlers.

  • Argument intrusiveness may now be zero. Previously it had to be a strictly positive value.

  • Add without_progress() - which causes all progression conditions to be muffled and ignored.

Bug Fixes

  • Progressor functions could produce progression conditions that had the same identifiers and therefore would be considered duplicates such that progression handlers would ignore them.

  • It was an error if a progression took a step big enough to skip more than the next milestone.

  • Progression handlers now keep the internal step field within [0, max_steps] in case of a too big progression step is taken.

  • Progression updates received after progression handler is finished would keep increasing the internal step field.

Significant Changes

  • Renamed restart consume_progression to muffleProgression to align with restarts muffleMessage and muffleWarning in base R.

New Features

  • Add a plyr-compatible “progress bar” named progress_progressr().

  • Add option progressr.clear.

  • Visual progression handler will now always render the complete update state when clear is FALSE.

  • Now progression handlers ignore a re-signaled progression condition if it has already been processed previously.

  • Now each progression condition holds unique identifiers for the R session and for the progressor that produced the condition. It also contains an unique index per progressor that is incremented whenever a new progression condition is created.

Significant Changes

  • First decent prototype of this package and the idea behind it.

  • Make auto_done = TRUE the default.

New Features

  • Add argument auto_done to automatically have progress updates also signal “done” as soon as the last step has been reached.

  • Made amount the first argument of progressors to avoid having to specify it by name if progressing with an amount than the default amount = 1.0.

  • Add argument clear to control whether progress reporter should clear its output upon completion. The default is to do this, where supported.

  • Add progress update handler based on pbmcapply::progressBar().

  • Each achieved step is now timestamped.

  • Add option progressr.debug.

New Features

  • Add intrusiveness parameter that specifies how intrusive/disruptive a certain progress reporter is. For instance, an auditory reporter is relatively more disruptive than a visual progress bar part of the status bar.

  • Simplified the API for creating new types of progress reporters.

New Features

New Features

New Features