gsoc'20 call summary

[tdhock]

TODO post summary of calls.

[tdhock]

• push/commit more often (at least once per day).
• send me links to blog post to read.
• currently upgrading testing packages. some issues. clickElement function not working. tried to execute javascript using executeScript works. change implementation of clickID? report as an issue to RSelenium? Minimal reproducible example. double check if the ID is there and only occurs once -- sometimes selenium does not work when there is more than one element with the same ID.
• Rmd rendering? calling the knit print test, it renders the page, but instead of the animations, its just showing what the animint package shows on the console. look in the DOM? can't find the animint.js file? based on output looks like print method is being run, but knit_print should be run, maybe look at knitr / rmarkdown to see if there are any new ways to handle knit_print methods? https://cran.r-project.org/web/packages/knitr/vignettes/knit_print.html

> methods(knitr::knit_print)
 [1] knit_print.animint                    knit_print.data.frame*               
 [3] knit_print.default*                   knit_print.grouped_df*               
 [5] knit_print.html*                      knit_print.knit_asis*                
 [7] knit_print.knitr_kable*               knit_print.pillar_squeezed_colonnade*
 [9] knit_print.rowwise_df*                knit_print.shiny.tag*                
[11] knit_print.shiny.tag.list*            knit_print.tbl_sql*                  
see '?methods' for accessing help and source code
> 

• moving testing to docker? make a blog post or screencast to explain how to do that.
• oop references: look at animint2/R/geom-*.r files and help(gganimintproto)

[lazycipher]

• Why GSoC?
• Push regularly. The pushed code needs not necessarily to be perfectly working.
• Consolidate clickHTML and clickID if possible.
• Report dependency issues to the developers with minimal reproducible examples.
• knit_print.Rmd rendering due to print method issues.
• Moving tests to docker and write a post about that and if possible make a video.
• OPPs reference with help(gganimintproto) see implementation in geom- files.
• Write every week or every 2 weeks a blog post.

[tdhock]

hi @lazycipher would you like another call next week?
if not please mention me in any issues you would like me to help with.

[lazycipher]

hi @tdhock, Can we have a call?
I'm having issues in understanding what causes this issue https://travis-ci.org/github/lazycipher/animint2/jobs/702589867#L9147
and a few more things.

[tdhock]

to solve that issue just delete that example and then recompile documentation via devtools::document()

> d + stat_sum_df("mean_sdl", mapping = aes(group = cyl))
Don't know how to automatically pick scale for object of type quosure/formula. Defaulting to continuous.
Don't know how to automatically pick scale for object of type quosure/formula. Defaulting to continuous.
Error: Aesthetics must be either length 1 or the same as the data (32): x, y

[tdhock]

next week is really not that great for me for a call so let's try to use github ok?

[lazycipher]

@tdhock sounds good to me!
I'll recompile and the push. I missed that part.

[lazycipher]

Hii @tdhock, Can you check what's this test renders and probably explain this test a bit to me, please?
https://github.com/tdhock/animint2/blob/master/tests/testthat/test-renderer4-Inf.R

The test is failing because of line 36.

[tdhock]

this test is supposed to make sure that Inf/-Inf are rendered properly on both the x and y axes (at the edge of the panel).
the line number 36 is the following which computes the y position of the bottom of one of the drawn line segments

seg.bottom <- as.numeric(attr.mat["y1",])

[lazycipher]

@tdhock, whenever you get free, I need your help to solve the above issue and #40.

[tdhock]

hi what is the above issue? looks like the viz with Inf/-Inf is rendering fine so you just need to make sure the testing code is extracting the positions correctly right?

[lazycipher]

If that's the case, then we are looking in the wrong panels for lines.
I'll rectify this asap.

[lazycipher]

@tdhock, I don't know why it's happening! lazycipher@78c8535
The whole file is being taken as modified while when I commit, I check and I see only a single line being modified!
Update: Looks like those files had the CRLF line ends instead of LF.

[lazycipher]

The graph isn't rendering properly!
On R v4:

While on R v3.6.3:

[lazycipher]

Fixed this! It was stringsAsFactors = TRUE that was missing!

[tdhock]

great, that makes sense.

[lazycipher]

Hi @tdhock, @faizan-khan-iit, I'm having issues in getting the flow of data.
If possible can we have a call as soon as possible please so that I can proceed with #43?

I've also made and posted 2 videos of testing on YouTube which shows how it's done on Linux and Windows. Soon I'll be posting a few videos in which I'll be making animated viz.

[lazycipher]

Call on 29th July with @faizan-khan-iit,

• Everything looks good till now: @faizan-khan-iit.
• I'm working on OOPs implementation. Things look quite good as of now. We should wait for @tdhock for his thoughts over the current implementation.
• A discussion about possible ways to make it work.
• Is it feasible to compute all required data from inside the geom_* functions?
• Once done with current OOPs implementation, we should go ahead with moving the calculation of subset
  from compiler to renderer.

@faizan-khan-iit, Please let me know if I missed anything.

Thank you so much for the detailed discussion.

[tdhock]

problem understand what is R package? no, problem understanding how R package works. that is normal.
details about problems?
move geom pre-processing to specific geom classes. issue is how? move common processing to base geom, then move geom-specific code to specific classes. sounds great, reasonable.
in base Geom class, you have the save/export method, then from within that method, you call the pre_process method. If l is the layer object then you call l$pre_process(). in base Geom class you need a pre_process method that is a no-op.
How do you get the modified data back? save the return value?