Open jernestmyers opened 1 year ago
We've had recent discussions about how the differences between our dev environments can show "false" problems, like the styling of the Featured variables in the viz var selections and the download modal's table headers that take 100% of the vertical height. I call these "false" problems because they only exist in the web-eda dev server, but not in the sites dev environments (ie, yarn nx start @veupathdb/clinepi-site).
These particular examples turned out to be due to the use of emotion's css prop in eda
without the proper babel config to interpret these styles. This PR adds the necessary babel config to handle the css prop.
This means the remaining issue boils down to the layout differences between eda site servers vs eda local server, which could be resolved by option 3, described above.
I think we should make sure we are using the same basic layout structure for all websites (i.e., libs
and sites
local dev sites). I think this is basically what you are saying. We might even want to make a new package called something like app-shell
, which includes the sticky header/footer behavior, etc; and it used by all websites.
In responding to this issue re: filter chip placements, my first PR failed QA. The issue was that I was only developing and testing in a
web-eda
server (ie,yarn nx start @veupathdb/eda
from root oryarn start
from thepackages/libs/eda
directory), which does not include theposition: sticky
site headers and thus I didn't account for their vertical space.We've had recent discussions about how the differences between our dev environments can show "false" problems, like the styling of the
Featured variables
in the viz var selections and the download modal's table headers that take 100% of the vertical height. I call these "false" problems because they only exist in theweb-eda
dev server, but not in thesites
dev environments (ie,yarn nx start @veupathdb/clinepi-site
).My QA failure is due to the opposite problem: a "false" positive, if you will. Working in
web-eda
gave me the false impression that my work was working as expected.Some (quick, non-exhaustive) ideas that come to mind that may help us get to a better state:
web-eda
server, then we must also "test" our changes in the affectedsites
server as well (most likely bothmbio-site
andclinepi-site
)enforce the same CSS in(see comment below)web-eda
as is enforced inmbio-site
andclinepi-site
web-eda
as necessary to better emulate our sites