@yakra: Thanks for the additional explanation. Sorry for the delay; the last month has been unusually busy for me.
First, I'm disappointed that you think I was unaware of HDX or how to use it properly. I am a frequent user of HDX and fully realize how it seems to indicate a problem here, hence the topic.
I-40 and I-55 are not actually concurrent here. I-55 exit 7 is within the footprint of the I-40/55 split, but the point where I-55 and AR 77 cross is located to the east of the center of the split. The I-40/I-55 point is where it needs to be according to the
CHM guidelines, as is the I-55/AR 77 point. It is messy because reality is messy here.
I acknowledge that you disagree that graph connections need to exist here. CHM guidelines were written prior to Jim's assumption of the project, and collaborators have occasionally mentioned Tim's insistence that graph connections were not a priority. However, Jim has indicated on numerous occasions that graph connections are helpful for the METAL project, and has demonstrated his interest in this by creating tools for us to use to identify NMPs and other errors. The fact that he refers to them as errors tells me that minimizing them should be a priority. Granted, there are cases where graph connections can't be made for one reason or another. It's my opinion that this is not one of those cases. The Texas examples you bring up are different (lack of direct access, or different exit numbers from opposite directions). They are much more like I-40 exit 276 (from eastbound I-40 to AR 77 from the opposite side of the split), which correctly does not have a graph connection to AR 77 since access is made via a long stretch of frontage road. Nevertheless, even if your examples were identical to this one, I would be comfortable with how you handled them because they are consistent with the guidelines, such as they are.
(Regarding your TXSpr93 example, it seems that the US69 point would be more properly labeled US69/287 or US69/96.)
Concurrencies certainly are important. However, I fail to see how this could be "messing with people's stats" any more than the alternatives.
Example: a user who has never travelled on I-40 takes I-55 southbound to exit 7 and continues on AR 77. As things are set up currently, she would get credited for mileage on I-55 and AR 77 but not I-40, with the benefit of having a link to AR 77 show up from the I-55 page in the HB. She would not receive any credit for travelling on I-40.
If Exit 7 were eliminated, the user would either have a gap in her maps and a few tenths of a mile less of I-55 than she had actually travelled, or might instead feel the need to claim mileage all the way to the next waypoint south (I-40 exit 278). Either way, that would mess with her stats, in my opinion. Likewise, if Exit 7 was added as a hidden point on I-40, the segment of I-55 between waypoints 7 and 8 would also count as mileage on I-40, which she had technically not travelled. The concurrency detector would not give her any additional mileage, but she would have a short distance of I-40 showing up in her table, which would also mess with her stats.
In summary, I believe that the slight increase of mileage in Arkansas caused by not treating I-40 and I-55 as concurrent between I-40 points 277 and 278 is no more incorrect than the slight decrease of eliminating the point for the I-55/AR 77 junction. As a result, I am leaving things as they are here.
