Environment
- treepplr 0.14.0
- tpplc 0.4
Issue
align is a documented compile option name in treepplr:::tpplcCompileOptions, so it's natural to write:
tp_compile(model_path, method = "smc-bpf", align = TRUE)
(once #17's unlist() bug is worked around, this still fails:)
Unexpected argument: '--align' is not valid after '--method smc-bpf'
Per tpplc --help, alignment for smc-bpf/smc-apf is instead requested via --resample align (the help text for --resample notes the align value "forces --align" internally) — a bare --align flag isn't in either method's usage line at all.
This might be intentional CLI design (maybe --align is only meaningful standalone for other methods, e.g. mcmc), in which case this is purely a documentation/validation gap rather than a compiler bug. Either way, nothing in tp_compile() or its documentation currently explains this, so a user has to reach for tpplc --help themselves and reverse-engineer the correct combination by trial and error.
Suggestion
- Document (in
tp_compile()'s help, and/or in a vignette on SMC methods) that alignment for smc-bpf/smc-apf requires resample = "align", not a bare align = TRUE.
- Optionally, validate incompatible option combinations before invoking
tpplc and raise a clearer R-level error.
Environment
Issue
alignis a documented compile option name intreepplr:::tpplcCompileOptions, so it's natural to write:(once #17's
unlist()bug is worked around, this still fails:)Per
tpplc --help, alignment forsmc-bpf/smc-apfis instead requested via--resample align(the help text for--resamplenotes thealignvalue "forces--align" internally) — a bare--alignflag isn't in either method's usage line at all.This might be intentional CLI design (maybe
--alignis only meaningful standalone for other methods, e.g.mcmc), in which case this is purely a documentation/validation gap rather than a compiler bug. Either way, nothing intp_compile()or its documentation currently explains this, so a user has to reach fortpplc --helpthemselves and reverse-engineer the correct combination by trial and error.Suggestion
tp_compile()'s help, and/or in a vignette on SMC methods) that alignment forsmc-bpf/smc-apfrequiresresample = "align", not a barealign = TRUE.tpplcand raise a clearer R-level error.