DC Feed Confusion

[GoogleCodeExporter]

Dan, help us out here....

"Success" with no polling place  when there's no segment (that I can find)
422 RANDOLPH ST NW WASHINGTON DC 20011
http://pollinglocation.apis.google.com/?q=422%20RANDOLPH%20ST%20NW%20WASHINGTON%
20DC%2020011
{"status":"SUCCESS","locations":[[]],"contests":[[]]}

Same thing for when there is a segment
444 RANDOLPH ST NW WASHINGTON DC 20011

<street_segment id="65937">
    <start_house_number>444</start_house_number>
    <end_house_number>444</end_house_number>
    <odd_even_both>Even</odd_even_both>
    <non_house_address>
      <street_name>RANDOLPH</street_name>
      <street_suffix>ST</street_suffix>
      <address_direction>NW</address_direction>
      <city>WASHINGTON</city>
      <state>DC</state>
      <zip>20011</zip>
    </non_house_address>
    <precinct_id>400045</precinct_id>
    <precinct_split_id>4510</precinct_split_id>
  </street_segment>

But then for:
1830 Constitution Ave NW Washington DC
http://pollinglocation.apis.google.com/?q=1830%20Constitution%20Ave%20NW%20washi
ngton%20dc
{"status":"NO_STREET_SEGMENT_FOUND","locations":[[]],"contests":[[]]}

When:
  <street_segment id="10131">
    <start_house_number>1830</start_house_number>
    <end_house_number>1830</end_house_number>
    <odd_even_both>Even</odd_even_both>
    <non_house_address>

      <street_name>CONSTITUTION</street_name>
      <street_suffix>AVE</street_suffix>
      <address_direction>NE</address_direction>
      <city>WASHINGTON</city>
      <state>DC</state>
      <zip>20002</zip>
    </non_house_address>
    <precinct_id>400086</precinct_id>
    <precinct_split_id>8608</precinct_split_id>
  </street_segment>

Original issue reported on code.google.com by aaro...@gmail.com on 21 Oct 2010 at 2:29

[GoogleCodeExporter]

Looking into the first.
Note that the second has a segment corresponding to NE, not NW.
Maps believes NE is fine, so 
http://pollinglocation.apis.google.com/?q=1830%20Constitution%20Ave%20NE%20washi
ngton%20dc
works

Or rather, returns success.
i'm looking into the polling id issue.

Original comment by dan...@google.com on 21 Oct 2010 at 3:08

[GoogleCodeExporter]

The segments all reference precinct splits that don't exist in the file.

[dannyb@dannyb-macbookpro:~]> grep 12704 vipFeed-11.xml 
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>
    <precinct_split_id>12704</precinct_split_id>

While the lookup understands to look at precinct when precinct split does not 
define a polling location, it does not understand to look at precinct if the 
precinct split id is simply invalid :)

Original comment by dan...@google.com on 21 Oct 2010 at 4:29

[GoogleCodeExporter]

I'm not sure good error behavior would be to look at precinct, since that could 
give us bad polling locations.

Original comment by dan...@google.com on 21 Oct 2010 at 4:29

[GoogleCodeExporter]

Please try both ways. If a precinct link to polling location is working, use 
it. I give you permission to blame me later if this fails.

Original comment by aaro...@gmail.com on 21 Oct 2010 at 4:33

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Right now this doesn't happen because we take the precinct_id off the precinct 
split, not the original segment.

Is segment->precinct_id ever different than 
segment->precinct_split->precinct_id?

Original comment by dan...@google.com on 21 Oct 2010 at 5:18

[GoogleCodeExporter]

Only different if there's a programming screw-up like in DC...if one path fails 
you should try the other path before giving up.

Original comment by aaro...@gmail.com on 21 Oct 2010 at 5:30

[GoogleCodeExporter]

Okay, i'm fixing this then in the pipeline, but leaving it assigned to jared to 
fix the feed.
Should be live in an hour or two (it may wait till morning if i fall asleep on 
my keyboard)

Original comment by dan...@google.com on 21 Oct 2010 at 5:46

[GoogleCodeExporter]

Dan -- I still just see "SUCCESS" here..thoughts?

Original comment by aaro...@gmail.com on 21 Oct 2010 at 8:17