use OrderDict because csv.DictReader returns a dict after version 3.7

[andreas-stuerz]

See: python/cpython#8014

[fraenki]

LGTM, I'll apply the same fix for the statistics script.

[AdSchellevis]

I don't want to spoil the fun here, but are we sure we're fixing the right thing? The end result seems to be a dict() anyway

plugins/net/haproxy/src/opnsense/scripts/OPNsense/HAProxy/lib/haproxy/cmds.py

Line 333 in dbb697a

servers.append(dict(row))

[kulikov-a]

@AdSchellevis imho not just a dict() is needed, but an OrderedDict for move_to_end() to work at

plugins/net/haproxy/src/opnsense/scripts/OPNsense/HAProxy/lib/haproxy/cmds.py

Lines 329 to 331 in c149e55

	row['id'] = f"{row['pxname']}/{row['svname']}"
	row.move_to_end('id', last=False)
	servers.append(dict(row))

[AdSchellevis]

@kulikov-a it won't stick to the end when casting to dict() later on :) (if it does, it's pure luck)

[kulikov-a]

@AdSchellevis ah, got it, thanks. but shouldn't the order be preserved if (starting from 3.7) it is declared that the dict() preserves the insertion order (https://docs.python.org/3/whatsnew/3.7.html)?

[andreas-stuerz]

@AdSchellevis
This fixes the bug and don´t modifiy the existing structure as casting to a dict should preserve the order. But maybe me, @kulikov-a and the python guys are overlooking something? Do you have any other information?

[AdSchellevis]

@kulikov-a I kind of missed that item in the release notes, my point just is that if the order doesn't matter in real life, best don't try to make it look like it does..... trusting ordering in named array types is often quite dangerous and prone to errors. Eventually it wraps into a json container, which likely doesn't care about order either.

plugins/net/haproxy/src/opnsense/scripts/OPNsense/HAProxy/lib/haproxy/cmds.py

Line 87 in c149e55

return json.dumps(resObj)

It's not my code, if @fraenki choses to accept it I don't mind, spending years on reverse engineering of all sorts of weirdness in our old code just makes my eyes hurt.

[kulikov-a]

@AdSchellevis thanks for the explanation. thought that json.dump() just inherits the dict() behavior and now the order should be preserved for json too (https://docs.python.org/3.8/library/json.html).
@Andeman sorry for the discussion. just trying to learn something

[AdSchellevis]

@kulikov-a that is good to know, but since it wasn't earlier, its likely safe to assume order doesn't matter (in which case someone could just add "id" to the container and be done with it)

[fraenki]

I can confirm that it fixes the maintenance page, so the targeted issues is solved.