Inaccurate display of navigation curve

As I understand it, the only thing the FR255 really understands is "courses" which I think use the FIT file format.

So when create a course in Connect, you get this directly. The problem you experience sounds like an import/translation problem. You might look for options to import differently, or some external converter that can produce the FIT file for you and then upload that to Connect?

I've seen similar issues trying to exchange course info between Connect and Explore. I use Connect with my FR255 and Explore with my inReach MIni 2. It is a pain to try to get a good wilderness hiking course that I can use on both devices (i.e. for redundancy).

I make courses for the FR255 with Strava. Very nice editor.

I downloaded the FIT file saved in the COURSES folder on the watch and tried to re-import it as a new route in Garmin Connect - it looked undeformed, the same as the original source GPX file.
So I see the problem in the way the watch displays it. Supposedly, more powerful Garmin models are better.

The fact that it can display courses just fine tells me it is a problem with the initial data or the conversion process. Maybe the GPX points don't get translated into the right point type in the FIT file? Not all FIT files are equivalent. The FIT format is capable of storing points of different types, some defining the course line and others defining waypoints or adjacent landmarks.

Also, I think traditional GPX file usage may be sparse and assume additional routing during display. So it is key waypoints along the way but expects the display to fill in the gaps with knowledge of terrain or trail networks.

For a device like FR255, there is no supporting map. So it will only draw straight lines connecting the points. A good course is going to have a high resolution series of points quite close to each other to trace out the full shape of the route.

I guess I didn't express myself completely clearly: Displaying only points from the GPX file connected by straight lines would be enough for me, so I don't see the source GPX file as incorrect or insufficient, just like the FIT file generated from it. I don't see any problems with shorter routes or with displaying them on the watch, I just came across the described problem now. Czech Garmin support confirmed this problem to me, but they said they can't give details under what exact conditions it occurs.
That's why I'm trying to draw attention to this problem on this forum, because the watch manual is silent about it and Garmin Connect doesn't warn about route reduction in a specific case when preparing (importing) such a route, which I find annoying and in a sense even dangerous. Moreover, as a programmer, I don't really understand why the reduction has to occur.

FWIW, I downloaded my longest course (about 27 km) from Connect as a FIT file and used fitdump to look at it. This was one I created on Connect itself. This one worked nicely on my FR255 during the actual hike.

It seems like it has about 839 "record" entries which I think are track points. Then just a handful of "course_point" entries which I think were waypoints I labeled on the course editor like water crossings.

When I download the GPX version, it also has 839 "trkpt" records in it.

Luděk The behavior you're seeing may be related to course limitations on the device. Garmin watches have limits related to course points and course calculation, and long or complex courses may require splitting into smaller sections. If you're working with a very detailed course, reducing its complexity or dividing it into smaller segments may help avoid course-related limitations. 

Thank you for your response Garmin, Czech support also told me something similar. But I don't understand:
1. Why can't the watch correctly draw the navigation route when it can correctly draw the actual route (black line), which in my case was a little longer than planned and with more points (5553 compared to 1671)?
2. Why is Garmin Connect unable to announce when preparing the route whether it will be reduced by the watch (and possibly show how) so that the user knows whether he can rely on the navigation, which can be a matter of life or death?
3. Why doesn't Garmin Connect adjust the route in some more intelligent way (e.g. Douglas-Peucker algorithm) than the watch probably does with the sampling mentioned above? Garmin Connect, whether web-based or as a mobile application, generally has more power than the watch for this.
4. And if you can't or don't want to fix any of the previous points, why don't you at least publish the exact criteria for when the watch will reduce the route so that the user can take this into account and possibly adjust the route themselves?

This would also correspond to my observations - I don't see any problems with navigation with the watch on shorter routes (either the route doesn't deform at all or only slightly). But on the mentioned route it was already clearly visible in several places. It's a shame that the GPX or FIT file in question cannot be attached here - I don't understand this limitation here either.

Garmin, can you please respond to the 4 points of my response, or at least the last one?