pb_tick(self, private, len, tokens) : !self$finished is not TRUE

[bokov]

ReprEx:

try(rm(pb0));
pb0 <- progress_bar$new(format = "  got :current rows at :tick_rate/sec, total time :elapsed, total rows :total", clear = FALSE);
for(ii in 1:10) {Sys.sleep(1/10);pb0$tick(1000)};
pb0$terminate();

Expected behavior:

# got 10000 rows at 10989/sec, total time  1s

Observed behavior:

# Error in pb_tick(self, private, len, tokens) : !self$finished is not TRUE

Workaround:

try(rm(pb0));
pb0 <- progress_bar$new(format = "  got :current rows at :tick_rate/sec, total time :elapsed", clear = FALSE, total=NA);
for(ii in 1:10) {Sys.sleep(1/10);pb0$tick(1000)};
pb0$terminate();

Both changes are necessary-- removing any reference to :total when the total is unknown and adding an explicit total=NA.

I think that when the tick-count exceeds the default value for total (100), this error is triggered unless total is explicitly set to NA.

Yes, it's on me for failing to register that :total is redundant with :current when the total is unknown. But I see a number of downstream unresolved problems featuring this error over the years:

mikajoh/stmprinter#2
r-dbi/bigrquery#269
https://stackoverflow.com/questions/52655172/r-cannot-import-data-through-bigrquery-package-selffinished-is-not-true

So perhaps there would be a benefit to making this more obvious. Some possibilities:

• Explicitly saying in the help file that the tick counts must not be permitted to exceed the expected total
• Catching and handling that case-- e.g. by letting the progress bar to seem to go beyond its expected limit as some apps do.
• Replacing the current !self$finished is not TRUE message with a more informative one.

Until then, hopefully this post will help whomever else has this problem and is persistent enough to follow it here. :-)