Refactoring merged in #1284 wasn't perfect and left per-image .csv files being written to
<output>/<filename>/processed whilst all other per-image files were written to <output>/processed/<filename> (see
#1290 for the bug report).
This was resolved in #1291 along with including a fix for #1287 but in discussion it was
questioned as to the utility of retaining the
feature of writing statistics to individual files.
I do wonder though, as I had this problem when working on #1284, what the utility of duplicating output is. The all
statistics .csv files at the top level of the output_dir contain the same results as the per-image, it just requires
filtering. Maintaining such convenience functionality detracts from developer time to work on other tasks.
Basic data manipulation skills are invaluable and something all scientific researchers should have the in their skill
set. An alternative if a single images output really is required is to just re-run TopoStats on just the single file.
It was noted by @tcatley that they don't use the per-image statistics and instead just filters from the "all statistics"
set of results, although this may not be the case for everyone.
In general though maintaining such convenience functions does take development time away from working on other aspects
of code development, and so it is proposed to remove io.save_image_grainstats(), its associated tests and the code
that calls it so that the code-base is slightly smaller and there is less maintenance overhead.
Refactoring merged in #1284 wasn't perfect and left per-image
.csvfiles being written to<output>/<filename>/processedwhilst all other per-image files were written to<output>/processed/<filename>(see#1290 for the bug report).
This was resolved in #1291 along with including a fix for #1287 but in discussion it was
questioned as to the utility of retaining the
feature of writing statistics to individual files.
It was noted by @tcatley that they don't use the per-image statistics and instead just filters from the "all statistics"
set of results, although this may not be the case for everyone.
In general though maintaining such convenience functions does take development time away from working on other aspects
of code development, and so it is proposed to remove
io.save_image_grainstats(), its associated tests and the codethat calls it so that the code-base is slightly smaller and there is less maintenance overhead.