You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In gh-927 I tried to convince the Argonne team, unsuccessfully, to add unit tests for the appearance (image grid-ordering) of the HTML reports produced by the python -m darshan summary .. code. The suitability of CSS in-browser rendering for unit testing has been debated more broadly in places like here: https://stackoverflow.com/questions/342579/css-unit-testing
I think the main argument against the testing is just the likely-large time investment needed. On the other hand, I'd much prefer to have unit tests enforce that common browsers are displaying what the team wants displayed instead of burdening i.e., scientific Python code reviewers with assessment of CSS layout interactions, and basically mandating more visual inspection on complex PRs since we can't trust the static HTML ordering to be what the browser renders as the CSS grows in complexity.
Another useful alternative may be to have a team member who is an expert at CSS/web design/layout testing deal with the code review aspects related to this, but this strikes me as unlikely anytime soon.
The text was updated successfully, but these errors were encountered:
In gh-927 I tried to convince the Argonne team, unsuccessfully, to add unit tests for the appearance (image grid-ordering) of the HTML reports produced by the
python -m darshan summary ..
code. The suitability of CSS in-browser rendering for unit testing has been debated more broadly in places like here: https://stackoverflow.com/questions/342579/css-unit-testingIn some of those discussions, CSS unit testing frameworks like https://github.com/jamesshore/quixote have been brought up.
I think the main argument against the testing is just the likely-large time investment needed. On the other hand, I'd much prefer to have unit tests enforce that common browsers are displaying what the team wants displayed instead of burdening i.e., scientific Python code reviewers with assessment of CSS layout interactions, and basically mandating more visual inspection on complex PRs since we can't trust the static HTML ordering to be what the browser renders as the CSS grows in complexity.
Another useful alternative may be to have a team member who is an expert at CSS/web design/layout testing deal with the code review aspects related to this, but this strikes me as unlikely anytime soon.
The text was updated successfully, but these errors were encountered: