'Not a promise' error

[davidlockhart]

I have a panel dataset for which I have created lagged variables using the lag() function. When I try to calculate the delta for each timepoint, using the mutate command below, I get the error message "Error: not a promise".

> kw.lags[,c("imps", "lag1_imps", "lag2_imps")]
Source: local data frame [157,737 x 3]
Groups: 

   imps lag1_imps lag2_imps
1    65        NA        NA
2    79        65        NA
3    62        79        65
4    69        62        79
5     1        NA        NA
6     2        NA        NA
7     2         2        NA
8     1         2         2
9     2         1         2
10    5        NA        NA
..  ...       ...       ...

> kw.deltas <- mutate(kw.lags, 
+   d1_imps = imps - lag1_imps,
+   d2_imps = imps - lag2_imps
)
Error: not a promise  

[hadley]

Could you please send a reproducible example? i.e. by including some code to reproduce kw.lags or a tbl that illustrates the same problem.

[romainfrancois]

... @davidlockhart do you have a reproducible example. We can't fix this otherwise as we can't reproduce the problem.

[davidlockhart]

After reporting the issue, I implemented workarounds and was subsequently unable to repro myself. I believe I must have accidentally done something unexpected to the kw.lags object. I loaded a saved data object from the morning the error occurred, but it is working fine as well.

Sorry for the false alarm, if this or something similar happens again, I will let you know.

[davidlockhart]

I now have a reproducible example of very similar behavior to issue #314.
Attached is an RData file with an object called profile_summs that
generates the same error ("Not a promise") when I run the following:

summarise(group_by(profile_summs, impression.chg,add=FALSE),
num_keywords = sum(n),
orders = round(mean(orders,na.rm = TRUE)/n,1),
gbv = round(mean(gbv,na.rm = TRUE)/n,1),
profit = round(mean(profit,na.rm= TRUE)/n,1),
cpc = round(mean(cpc,na.rm = TRUE),2)
)

(I have added a large amount of noise to the substantive data to maintain
confidentiality of the source data, but I confirmed that the error persists
with this data object.)

Here is the code that creates
profile_grouped <- group_by(week20.dt, profilemain,add=FALSE)
profile_summs <- summarise(profile_grouped,
impression.chg = mean(pctimp),
ctr.chg = mean(pctctr),
cvr.chg = mean(pctcompcvr2),
position.chg = mean(pctposn),
n = n(),
orders = round(sum(ordersmain,na.rm = TRUE)/n,1),
gbv = round(sum(gbv,na.rm = TRUE)/n,1),
profit = round(sum(gp_tracked_usd,na.rm=
TRUE)/n,1),
cpc = round(sum(cost)/sum(clicks),2)
)

profile_summs$impression.chg <- factor(profile_summs$impression.chg,levels
= c(0,1,2), labels = c("down","stable", "up"))
profile_summs$ctr.chg <- factor(profile_summs$ctr.chg,levels = c(0,1,2),
labels = c("down","stable", "up"))
profile_summs$cvr.chg <- factor(profile_summs$cvr.chg,levels = c(0,1,2),
labels = c("down","stable", "up"))
profile_summs$position.chg <- factor(profile_summs$position.chg,levels =
c(0,1,2), labels = c("down","stable", "up"))

Let me know if I can provide further assistance.

On Fri, Mar 14, 2014 at 11:04 AM, Hadley Wickham
notifications@github.comwrote:

Closed #314 #314.

Reply to this email directly or view it on GitHubhttps://github.com//issues/314
.

[dotcomken]

I received this error today in a similar way using summarise() with v0.1.3

It was a silly mistake on my part by having an extra comma like this:

summarise(x, a=min(aa), b=min(bb),)

Removed the extra comma and it worked fine.

[davidlockhart]

Thanks, Ken. I just verified that there was such an extraneous comma in
both of the instances I've found so far, and removing it does indeed fix
the problem.

On Mon, Mar 24, 2014 at 11:28 AM, Ken Pendergast
notifications@github.comwrote:

I received this error today in a similar way using summarise()

It was a silly mistake on my part by having an extra comma like this:

summarise(x, a=min(aa), b=min(bb),)

Removed the extra comma and it worked fine.

Reply to this email directly or view it on GitHubhttps://github.com//issues/314#issuecomment-38482387
.

[romainfrancois]

I fixed the issue reported by @dotcomken.
@davidlockhart I don't get the attachment, can you come up with code to reproduce this, or email it to me.

[hadley]

Could we just silently ignore missing arguments?

[romainfrancois]

Sure. I'll do that instead of throwing an error. Still no luck with reproducing @davidlockhart issue. Did you get the attachment data by any chance ?

[hadley]

@romainfrancois I'd say it was probably just a trailing comma that he later removed.

[romainfrancois]

Ok. I'll close the issue once I ignore instead of error then.

[romainfrancois]

Alright, slightly less trivial than I thought. What I'll do is skip the named_dots call entirely and put some more logic into DataDots: the class that handles promises and environments from ...