[Bioc-devel] progress with BiocParallel::bpiterate

[Robert Castelo]

hi Martin,

thanks for the explanation, it makes, i got confused because i was using 
'bpiterate()' to parallelize calculations through the slices of rows of 
a matrix as you described in this post:

https://support.bioconductor.org/p/123104

where this was the recommended option for parallelizing calculations 
through rows of a matrix. in such a case, i know how many iterations i 
have to do because is the number of row-chunks of the matrix. is there 
any alternative to use 'bpiterate()' in such case or is there a way to 
pass to 'bpiterate()' the number of iterations so that progress could be 
shown in %?

robert.


On 27/11/2019 19:16, Martin Morgan wrote:
> The rationale is that bpiterate() is used when the number of iterations is unknown (hence the length of the progress bar indeterminate...), whereas bplapply() is a more natural candidate for iteration when the length of the iteration is known... is bpiterate() providing functionality (REDUCE, maybe? that bplapply() does not?
>
> Martin
>
> On 11/27/19, 1:11 PM, "Bioc-devel on behalf of Robert Castelo" <bioc-devel-bounces using r-project.org on behalf of robert.castelo using upf.edu> wrote:
>
>      hi,
>      
>      the 'BiocParallel::bpiterate()' method shows progress through the
>      following call:
>      
>      setMethod("bpiterate", c("ANY", "ANY", "SerialParam"),
>           function(ITER, FUN, ..., REDUCE, init, reduce.in.order = FALSE,
>               BPPARAM=bpparam())
>      {
>      [...]
>           progress <- .progress(active=bpprogressbar(BPPARAM), iterate=TRUE)
>      [...]
>      }
>      
>      where the argument 'iterate=TRUE' when calling '.progress()' leads this
>      function to show progress by a message
>      
>      "iteration: x"
>      
>      where 'x' is a number from 0 to 'ntasks' within its code. i'd like to
>      know whether there is a particular reason why 'bpiterate' should show
>      progress in such a way and whether it would be possible to parametrize
>      it to show progress in the same way other 'BiocParallel' methods do,
>      such as 'bplapply()', 'bpvec()', etc..., i.e., doing
>      
>      |==========                        | X%
>      
>      
>      thanks!!!
>      
>      robert.
>      
>      _______________________________________________
>      Bioc-devel using r-project.org mailing list
>      https://stat.ethz.ch/mailman/listinfo/bioc-devel
>