I wrote an R function, filterVariantCalls(), taking a bgzipped VCF as input, filtering variants with too many missing genotypes, and writing a new bgzipped VCF. Here is an example:

library(rutilstimflutre) # my personal R pkg, see below
library(IRanges) # see https://support.bioconductor.org/p/73957/#73962
filterVariantCalls(vcf.file="input.vcf.gz", genome="test", out.file="output.vcf", yieldSize=10000, max.prop.gt.na=50)

It starts well:

FilterRules of length 1
names(1): filterGtProp
starting filter
filtering 10000 records

But then I receive the following error:

Error in extractROWS(x, eval(filter, x)) :
error in evaluating the argument 'i' in selecting a method for function 'extractROWS': Error in rule(envir) :
promise already under evaluation: recursive default argument reference or earlier problems?

Following R debugging good practices, I execute the traceback() function, which gives:

8: extractROWS(x, eval(filter, x))
7: .local(x, filter, ...)
6: subsetByFilter(vcfChunk, filters)
5: subsetByFilter(vcfChunk, filters)
4: .filter(file, genome, destination, verbose, filters, param, ...)
3: VariantAnnotation::filterVcf(file = tabix.file, genome = genome,
       destination = out.file, index = TRUE, verbose = (verbose >
           0), filters = filters)
2: VariantAnnotation::filterVcf(file = tabix.file, genome = genome,
       destination = out.file, index = TRUE, verbose = (verbose >
           0), filters = filters)
1: filterVariantCalls(vcf.file = vcf.file, genome = "test", out.file = out.file,
       yieldSize = 10000, max.prop.gt.na = 50)

Does the error come from the fact that stacks 5 and 6 correspond to the same function?

FYI the code of the filterVariantCalls() function is in my custom R pkg on GitHub. Note that, as this is my personal R pkg, it has many functions doing different stuff. To avoid having it to depend on too many packages from different fields (genomics, genetics, statistics, etc), I explicitly use requireNamespace("foo") when I need a function from the foo pkg in one of my function, as advised by Hadley Wickham in his book here.

Here is my sessionInfo():

R version 3.2.2 (2015-08-14)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS release 6.6 (Final)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats4    parallel  stats     graphics  grDevices utils     datasets
[8] methods   base     

other attached packages:
[1] IRanges_2.4.0          S4Vectors_0.8.0        BiocGenerics_0.16.0   
[4] rutilstimflutre_0.16.0

loaded via a namespace (and not attached):
 [1] AnnotationDbi_1.32.0       XVector_0.10.0            
 [3] GenomicAlignments_1.6.0    GenomicRanges_1.22.0      
 [5] zlibbioc_1.16.0            BiocParallel_1.4.0        
 [7] BSgenome_1.38.0            GenomeInfoDb_1.6.0        
 [9] tools_3.2.2                SummarizedExperiment_1.0.0
[11] Biobase_2.30.0             data.table_1.9.6          
[13] DBI_0.3.1                  lambda.r_1.1.7            
[15] futile.logger_1.4.1        rtracklayer_1.30.0        
[17] futile.options_1.0.0       bitops_1.0-6              
[19] RCurl_1.95-4.7             biomaRt_2.26.0            
[21] RSQLite_1.0.0              compiler_3.2.2            
[23] GenomicFeatures_1.22.0     Rsamtools_1.22.0          
[25] Biostrings_2.38.0          XML_3.98-1.3              
[27] VariantAnnotation_1.16.0   chron_2.3-47              

Pretty sure this is because your filter function filterGt has an argument max.prop.gt.na that references itself as its default. I will work on making the error message more informative in the future.