Make SMBs and predictions fit better

[scottleibrand]

This removes the leading 0 from boluses < 1U to make the labels fit better on closely spaced SMBs. It also adds a 2HR view, and limits the focus area to only display as many hours of purple prediction lines as the selected focusHours. This makes it possible to "zoom in" more on a period of interest.

[philipgo]

I think this is a great enhancement for OpenAPS users. Limiting the forecast range is a very good idea.

The font size is perfect on desktop computers, but a bit too small on my iPhone using the Nightscout app.

For me the tooltips around the SMB entries could be even smaller. With the current size in the rendering branch it is still impossible to hover over some BG entries. Alternatively, this might be solved by positioning the tooltip target area for Bolus/Carb entries under the symbol (where the caption is) and for BG entries over the dot.

[scottleibrand]

Is the font still too small after I bumped it up from 25 (in ZT) to 35 (in rendering)? I could try going back up to 40 (what's in dev) but that might start SMBs overlapping again, at least on anything longer than the 2h graph.

In the ZT branch I already narrowed the tooltip targets down to basically be the size of the dot plus the label (vs. having it be like 30m wide in dev). If you have a BG that's underneath a treatment dot, it won't be hoverable, but I don't think we want to try to "fix" that. We're already using the space above carbs for their labels and tooltip targets, and the space below insulin treatments for theirs. Not sure if there's any further tweaks that would be worthwhile, but if so you can play around with them at https://github.com/nightscout/cgm-remote-monitor/pull/2919/files#diff-de7f2411d8429f79b8d3436ca320c9dfR417 and PR something in if you find a better solution.

[scottleibrand]

sorry, I forgot to actually push fb4e74e until just now. @drnoname82 can you try that?

[philipgo]

I think the larger font size is not an improvement because it leads to overlapping and makes things harder to read (see screenshots). I would rather accept the smaller font size. Maybe a different font could lead to an even better result, but either way legibility is much better than in current dev.

I will play around with the tooltip targets. Probably this is not too important, but I would find it interesting to be able to delete single BG measurements much like other entries can be deleted by clicking and pulling them. This can only work if BG dots are not covered by tooltip targets.

[scottleibrand]

In playing around with label placement (which led nowhere useful) I discovered that we weren't properly scaling the size of treatment bubbles. This PR now allows larger insulin-only and carb-only bubbles for boluses >1U and carb entries >20g (using the default CR). It also switches to using the profile CR if there's no treatment-specific CR (before defaulting to 20, as before if there's no profile). And since that made boluses half the size they were before (now using a CR of 10 instead of the default 20), I doubled the size (area) of the bubbles so that 0.1U microboluses are the same size as BG dots, which makes large meals and boluses quite a bit more visible (at the expense of covering up more, unless you use the 2H view). Thoughts?

[diabeticdude2]

looks great to me. I am able to see the info I need. I have tested larger fonts, (40) but when I use that the numbers start to overlap. 30 is a bit too small, 35 seems like the happy middle place and is readable on my phone.

[philipgo]

This is very nice. I would still vote for the 25 font (or a better solution in case anyone has an idea) because at 30 SMBs given in two consecutive loops cannot be deciphered.

[scottleibrand]

There will always be overlap at some zoom levels unless we figure out a new way to do labels. The question is how far you have to zoom in to see them all clearly. It should now be possible to read all the labels without overlap at least at the 2H zoom level. Is that sufficient?

I hope 35 is sufficiently large that non-OpenAPS users won't see a problem with it. If it is, we might have to figure out how to dynamically scale the font based on the number of treatments displayed.

[garykidd]

I like 25...but to each their own I suppose. Might also consider 1 hour interval....or 90 mins even if you are having issues with overlap.

[scottleibrand]

Don't try 1.5h, or you'll have to find the variable in the Chrome debugger to change it back to a whole number to unbreak your NS. :-)

[cameronrenwick]

yup.. love it

[PieterGit]

Made a start in updating this branch to Node 8.8.1 and fixing the tests.
PR #2991 has now been merged to this PR.

[scottleibrand]

Now that the tests are fixed, what else needs to be done on this branch to get it into a suitable state for merge to dev?

[sulkaharo]

Looks good to go but this now targets ZT, not dev? :)

[scottleibrand]

Fixed.