journeys(): Walks have wrong distance

[dabund24]

I've noticed that legs that are walks have an incorrect distance value. The value is generally about 500 meters greater than the one displayed in DB Navigator, but the offset unfortunately doesn't really seem follow a pattern. Here are some examples:

from	to	DB Navigator	hafas-client	difference
Berlin Hbf (tief)	Berlin Hbf (S-Bahn)	80	500	420
Hamburg Hbf	Hauptbahnhof Süd, Hamburg	243	666	423
Köln Messe/Deutz	Köln Messe/Deutz Gl.11-12	59	615	556
München Hbf (tief)	München Hbf	130	692	562

This problem seems to be there on a lot of/maybe all profiles (I haven't tried all). I noticed it with the db profile. Options like a different walking speed or transfer time don't affect this.

[derhuerst]

(I looked into this using the DB profile.) It seems like, without opt.polylines (a.k.a. getPolyline), HAFAS returned wrong leg distances, presumably because they're just estimated or even a static fallback value.

With getPolyline: false:

{
    "type": "WALK",
    "dep": {
        "locX": 1,
        "dTimeS": "083100",
        "dTZOffset": 120,
        "type": "N"
    },
    "arr": {
        "locX": 3,
        "aTimeS": "084000",
        "aTZOffset": 120,
        "type": "N"
    },
    "gis": {
        "dist": 500,
        "durS": "000900",
        "dirGeo": 4,
        "ctx": "H|1|W$A=1@O=Berlin Hbf (tief)@L=8098160@a=128@$A=1@O=Berlin Hbf (S-Bahn)@L=8089021@a=128@$202309260831$202309260840$$$1$$$$$$|#VE#2#CF#100#CA#0#CM#0#SICT#0#AM#17#AM2#0#RT#8#",
        "gisPrvr": "H",
        "getDescr": true,
        "getPoly": true,
        "sumLDrawStyleX": 2,
        "resLDrawStyleX": 3,
        "prodX": 1
    },
    "resState": "N",
    "resRecommendation": "N"
}

With getPolyline: true:

{
    "type": "WALK",
    "dep": {
        "dInR": true,
        "dTZOffset": 120,
        "dTimeS": "083100",
        "locX": 17,
        "type": "N"
    },
    "arr": {
        "aOutR": true,
        "aTZOffset": 120,
        "aTimeS": "084000",
        "locX": 15,
        "type": "N"
    },
    "gis": {
        "ctx": "H|1|W$A=1@O=Berlin Hbf (tief)@L=8098160@a=128@$A=1@O=Berlin Hbf (S-Bahn)@L=8089021@a=128@$202309260831$202309260840$$$1$$$$$$|#VE#2#CF#100#CA#0#CM#0#SICT#0#AM#49#AM2#0#RT#7#",
        "dirGeo": 4,
        "dist": 80,
        "durS": "000900",
        "gisPrvr": "H",
        "poly": {
            "crdEncF": "??",
            "crdEncS": "NN",
            "crdEncYX": "q}q_IqbrpAfCmA",
            "crdEncZ": "??",
            "delta": true,
            "dim": 3,
            "ppLocRefL": [
                {"locX": 17, "ppIdx": 0},
                {"locX": 15, "ppIdx": 1}
            ],
            "type": "WGS84"
        }
    },
    "icoX": 11,
    "resRecommendation": "N",
    "resState": "N"
}

[derhuerst]

It seems like, without opt.polylines (a.k.a. getPolyline), HAFAS returns wrong leg distances […].

This doesn't seem to be the case with the VBB endpoint (and the parameters current VBB hafas-client profile, of course).

With getPolyline: false:

{
    "type": "WALK",
    "dep":
    {
        "locX": 1,
        "dTimeS": "083100",
        "dTimeFS":
        {
            "styleX": 0
        },
        "type": "N"
    },
    "arr":
    {
        "locX": 2,
        "aTimeS": "084000",
        "aTimeFS":
        {
            "styleX": 0
        },
        "type": "N"
    },
    "chg":
    {
        "txt": "Umstieg",
        "durS": "000900",
        "durFS":
        {
            "styleX": 1,
            "txt": "9 Min"
        }
    },
    "gis":
    {
        "dist": 67,
        "durS": "000900",
        "dirGeo": 10,
        "ctx": "H|1|W$A=1@O=S+U Berlin Hauptbahnhof (tief)@L=900003200@a=128@$A=1@O=S+U Berlin Hauptbahnhof@L=900003201@a=128@$202309260831$202309260840$$$1$$$$$$",
        "gisPrvr": "H",
        "getDescr": true,
        "getPoly": true,
        "sumLDrawStyleX": 2,
        "resLDrawStyleX": 3,
        "prodX": 6
    }
}

With getPolyline: true:

{
    "type": "WALK",
    "dep":
    {
        "locX": 1,
        "dTimeS": "083100",
        "dTimeFS":
        {
            "styleX": 0
        },
        "type": "N"
    },
    "arr":
    {
        "locX": 2,
        "aTimeS": "084000",
        "aTimeFS":
        {
            "styleX": 0
        },
        "type": "N"
    },
    "chg":
    {
        "txt": "Umstieg",
        "durS": "000900",
        "durFS":
        {
            "styleX": 1,
            "txt": "9 Min"
        }
    },
    "gis":
    {
        "dist": 67,
        "durS": "000900",
        "dirGeo": 10,
        "ctx": "H|1|W$A=1@O=S+U Berlin Hauptbahnhof (tief)@L=900003200@a=128@$A=1@O=S+U Berlin Hauptbahnhof@L=900003201@a=128@$202309260831$202309260840$$$1$$$$$$",
        "polyG":
        {
            "polyXL":
            [
                1
            ]
        },
        "gisPrvr": "H",
        "getDescr": true,
        "getPoly": true,
        "sumLDrawStyleX": 2,
        "resLDrawStyleX": 3,
        "prodX": 6
    }
}

[derhuerst]

This problem seems to be there on a lot of/maybe all profiles (I haven't tried all).

Which ones did you witness the behaviour with?

[dabund24]

It looks like I was a little bit too pessimistic about that issue after also having had stumbled across it with sbahn-muenchen and mobil-nrw. I've tested more profiles and vrn is another one with faulty distances. On the other hand, with bvg, oebb, nahsh and vmt Hafas returns distances properly. Also, thanks for figuring out the workaround, it fixes the issue for every tested profile :D