COMBINE_CON_RTES (DataProcessing)

[yakra]

Closes#492.

There will be a Web component to accompany this.

[jteresco]

Should I use this tonight?

[yakra]

Any thoughts on TravelMapping/Web#800?

[jteresco]

I think we can hold off until I can get you set up on a test server.

[yakra]

OK, sounds good.
In that case using this tonight sounds good.

You're off traveling now, right?
I might be able to set up a new server on one of my other lab machines. (Though I don't remember what I did back in 2019 and would have to mostly start from scratch again; may take a bit.)

Or stick bring lab2 back from the dead with a combination of its hardware and some other R420s I have.

Or -- hardware? -- plop its hard disk in an external adapter and boot lab6 up with it.
Yes. This may be my best bet for a quick solution? Nope.

[jteresco]

So if I give you write access on noreaster to /home/www/tmtest so you can put your Web files there, and we point it at the TravelMappingTest DB, which you should be able to populate with your version of the .sql file, would that test what you want?

[yakra]

Yeah, that'd work, thanks!
Not sure how I'd want to read into what db specifically; can I leave that to you?
the file is at /home/yakra/ytm/DataProcessing/siteupdate/cplusplus/TravelMapping.sql
Edit: Deleted it. I'll let you know when it's back.

(As an aside, if you're interested in seeing the latest bleeding edge cpp, go there and run git diff)

[jteresco]

You should have enough write access to put the version you want in /home/www/tmtest.

I can load up the DB when there's a file ready to go.

[yakra]

It's back.
(Ack, that took longer than I expected.)

[yakra]

just realized; it has no ranks table.
will see about appending
Done.

[yakra]

https://tmtest.teresco.org/devel/datacheck.php

[jteresco]

So ready for all this to go live tonight or is it better to work with it on tmtest for a bit first? It's easy enough to load up the DB.

[yakra]

OK, I've checked out everything I want to for hidden systems.

For real this time:

New DB, same path: /home/yakra/ytm/DataProcessing/siteupdate/cplusplus/TravelMapping.sql
Just the changes concerning this PR this time.
Let me know when you have it in and I can scp the .php files over.

[jteresco]

Reloaded the TravelMappingTest DB (sorry for the delay).

[yakra]

Looks good!

[yakra]

Since this isn't merged yet...
route_integrity.cpp does 7 different things. It kinda needs to be refactored.
Might do a force push. I'll let you know.

[yakra]

Good thing we held off on the merge.

Multi-threaded results are non-deterministic

7057d7056< indka.ka010;KA/TG;;;COMBINE_CON_ROUTES;indtg.ka010@KA/TG7823a7823> indtg.ka010;KA/TG;;;COMBINE_CON_ROUTES;indka.ka010@KA/TG

due to comparison of Waypoint addresses:

DataProcessing/siteupdate/cplusplus/classes/HighwaySystem/route_integrity.cpp

Line 118 in a93532f

if (w->route->region != p->route->region && w < p && this == p->route->system && &cr != cr2)

Solution: drop the w < p condition and compare the region pointers instead. Better results AND more efficient.
Better yet, compare the ConnectedRoute addresses.
Sure, the 1st option would short-circuit more efficiently, but I want to have the order of the links on the datacheck page row match the order of the routes in the _con.csv.

[yakra]

Force pushed & ready to rock!

route_integrity.cpp does 7 different things. It kinda needs to be refactored.

59% of it's gone compared to b046a77; 63% compared to a93532f.
Holding off on further refactoring until implementing hidden systems.
At that point it may even cease to exist as its own translation unit.

[yakra]

Let's hold off on this again while I read about git internals

[yakra]

Reverted to the 1st force push (2 commits refactoring + 1 for the new feature) from the 2nd force push (1 commit squashing it all together).
For a minute there I was hung up on making 2 or 3 blobobjects when I could be making 1, but meh, that's what pack files are for.
Commits are and are not diffs.
I'll err on the side of an understandable commit history, with discrete commits for discrete changes.

Ready to rock.

[yakra]

Hold on, gonna mess around with translation units. :P

[yakra]

Noticed that 82b03c4 added a couple dependencies to ConnectedRoute.cpp, so I decided to break out ConnectedRoute/dataceck.cpp into its own file, including a smaller subset of the headers in each.
While this does mean greater combined compile time due to double-processing the shared headers, it's outweighed by more often only having to recompile only one of the files.
Of course, ConnectedRoute is small potatoes. The real time savings would be heavy-hitters like Waypoint.cpp. I may get to that in the future; low priority.

Ready to rock!

...That is what I honestly believe.
Though of course, I do keep saying that. ;)
LOL quick, merge it in before I start tinkering around with whatever else!

---

dead branches:

• 46a47c4, same as 6d2945f without the #include change in ConnectedRoute.h
• d0f1894, indentation before squash/rebase

[jteresco]

OK, finally a chance to bring this in.

[yakra]

Ha! Glad I got a chance to tweak it.
I don't think I've anything left to improve here, so no regrets! :)