Weird "Non-continuous variable supplied to scale_x_continuous Error" - r

Question

I have an interesting conundrum. I can create the type of chart I seek interactively, but not automatically. Or, I nearly had it automatically, but something broke. (example data at end of post).
I have my loop working the way I would like, but have run into errors when I add some geom_vline() statements (for us, denoting significant changes in our production environment). I've tried working through it outside of the loop and am able to recreate the issue with details below.
I have the following steps:
create a vector with the list of changes:
changeVector <- c(as.Date("2011-11-30"),as.Date("2011-12-05"))
[WORKS] create a plot with the data below, and it works:
ggplot(df,aes(x=OBSDATE,y=AVG_RESP))+geom_line(aes(group=REGION,color=REGION))
[WORKS] try to add the geom_vline(xintercept=c(15308,15313)), and it works (but only if the geom_vline is at the end):
ggplot(df,aes(x=OBSDATE,y=AVG_RESP))+geom_line(aes(group=REGION,color=REGION))+geom_vline(xintercept=c(15308,15313))
[FAIL] try to add the geom_vline(xintercept=changeVector) - I had problems with this for some reason, and had to add as.numeric to recognize the vector values properly:
ggplot(df,aes(x=OBSDATE,y=AVG_RESP))+geom_vline(xintercept=as.numeric(changeVector))+geom_line(aes(group=REGION,color=REGION))
When this step runs, I get the wonderfully useful error message:
Error: Non-continuous variable supplied to scale_x_continuous.
So, any ideas? If I try to add an aesthetic component to the geom_vline, I still make no progress. My desire was to have the geom_vline preceding the geom_line because the vline is context, not data.
Thank you for your help!
Here is a subset of the data (dataFile name df):
OBSDATE REGION COUNT AVG_RESP
2011-11-29 EMEA 293 4.430375
2011-11-30 EMEA 299 4.802876
2011-12-01 EMEA 292 4.362363
2011-12-02 EMEA 293 4.209829
2011-12-03 EMEA 294 4.262959
2011-12-04 EMEA 294 4.207959
2011-12-05 EMEA 293 4.172594
2011-12-06 EMEA 293 4.230887
2011-12-07 EMEA 298 4.259329
2011-12-08 EMEA 293 4.197645
2011-11-29 Americas 296 2.841182
2011-11-30 Americas 296 2.932196
2011-12-01 Americas 292 2.766438
2011-12-02 Americas 293 2.819556
2011-12-03 Americas 291 2.710584
2011-12-04 Americas 295 2.728407
2011-12-05 Americas 290 2.764310
2011-12-06 Americas 290 2.817483
2011-12-07 Americas 295 2.733864
2011-12-08 Americas 291 2.732405
2011-11-29 APAC 328 7.294024
2011-11-30 APAC 325 7.091046
2011-12-01 APAC 314 6.969236
2011-12-02 APAC 327 6.920428
2011-12-03 APAC 325 7.226308
2011-12-04 APAC 324 7.046296
2011-12-05 APAC 318 7.075094
2011-12-06 APAC 317 7.016467
2011-12-07 APAC 318 7.187358
2011-12-08 APAC 318 7.310220

Answer 1

I'm not exactly sure why it is doing that, but here is a workaround that keeps the vertical lines behind the data lines:
ggplot(df,aes(x=OBSDATE,y=AVG_RESP)) +
geom_blank() +
geom_vline(xintercept=as.numeric(changeVector)) +
geom_line(aes(group=REGION,color=REGION))
EDIT:
Here is another workaround: explicitly specify that the x axis is to be a date, rather than have ggplot guess. When it guesses, it looks at the first layer plotted, which is the vertical lines. Given that the xintercept have to be given as numbers rather than dates, the x axis is assumed to be continuous/numeric. When the next layer is drawn, the dates of the x axis can not be mapped onto that and an error is thrown.
ggplot(df,aes(x=OBSDATE,y=AVG_RESP)) +
geom_vline(xintercept=as.numeric(changeVector)) +
geom_line(aes(group=REGION,color=REGION)) +
scale_x_date()