R evaluation of linear model: Error in table: all arguments must have the same length, how to solve?

I am new to R, and I am using the biopsy data set included with R, but I am encountering an error that says

Error in table(training$Direct.Deposit, predict.status) : all arguments must have the same length.

The issue arises when I run this line :

ct=table(eval$class, predict.status)

I tried troubleshooting but could not figure out the solution to this error message; any help will be appreciated! Here is my code:

library(MASS)
data("biopsy")

summary(biopsy)

#the thinnest clump is 1, while the largest is 10. The average clump thickness is 4.418.

sum(biopsy$class == "benign", na.rm=TRUE)
# 458 tumors are benign
458/699
#65.55% of the tumors are benign

sum(biopsy$class == "malignant", na.rm=TRUE)
# 241 tumors are malignant
241/699
#34.48% of the tumors are malignant



biopsy60train <- floor(0.6 * nrow(biopsy))
#419 observations
biopsy40eval <- floor(0.4 * nrow(biopsy))
#279 observations

training<-(sample(nrow(biopsy), nrow(biopsy)*.6))
training<-biopsy[training,]
#Training dataset 60%


eval<-(sample(nrow(biopsy), nrow(biopsy)*.4))
eval<-biopsy[eval,]
#eval dataset 40%


training$class<-ifelse(training$class=="benign",1,0)
training
#Convert class to binary (1,0)

#Fit logistic regression
training[1,]
m1=glm(class~V1+V2+V3+V4+V5+V6+V7+V8+V9, data=training, family="binomial")
summary(m1)

#p value of #9.14e-09 is small so we Reject the null hypothesis(ho), the relationship is significant.


eval$class<-ifelse(eval$class=="benign",1,0)
eval
#Convert class to binary (1,0)

#Fit logistic regression
eval[1,]
m2=glm(class~V1+V2+V3+V4+V5+V6+V7+V8+V9, data=eval, family="binomial")
summary(m2)

# Conclusion ---------------------------------------------------
# V2,V5, V7, V8, V9  are not significant, therefore can be removed from the analysis
 
m3=glm(class~V1+V3+V4+V6, data=eval, family="binomial")
summary(m3)
# Conclusion ---------------------------------------------------
# All independent variables are significant


# Next, we want to evaluate the model. To do that we should calculate 
# the TRUE POSITIVE RATE and the FALSE POSITIVE RATE

predict.prob=predict.glm(m3,type="response")
predict.status=1*(predict.prob>0.5)

ct=table(eval$class, predict.status)

TPR=ct[2,2]/sum(ct[2,])
FPR=ct[1,2]/(sum(ct[1,]))