wb_set_row_heights: replacement has N rows, data has N-1 at row 100000

[Arthfael]

I am writing large Excels with this package. As discussed with you here some time ago ( https://github.com/JanMarvin/openxlsx2/discussions ), I tested a parallel approach to building the table. I rewriting my old code, I encountered a weird error: I tried to set row_heights and the following call:
> WB <- openxlsx2::wb_set_row_heights(WB, SheetName, rws, 15) # rws = indices of the targe rows
failed for only one of my 55 parallel chunks, with an error of the type replacement has N rows, data has N-1 at row 100000
Downstream investigations revealed that this happened specifically when row 100000 was included in rws. More puzzling, functionally equivalent but syntactically different ways to pass rws make the difference between success and failure.

Below is a reprex:

WB <- wb_workbook()
WB <- wb_add_worksheet(WB, "test")
dat <- as.data.frame(matrix(rep(1, 10*2000), ncol = 10)) # Simple data to show that the problem is not caused by weird cell content
OS <- 99000
rws <- 1:2000 + OS
WB <- openxlsx2::wb_add_data_table(WB, "test", dat,
                                   dims = wb_dims(OS+1, 1),
                                   col_names = FALSE,
                                   table_style = "TableStyleMedium2",
                                   banded_rows = TRUE,
                                   banded_cols = FALSE)
WB <- openxlsx2::wb_set_row_heights(WB, "test", rws, 20) # -> Doesn't work
WB <- openxlsx2::wb_set_row_heights(WB, "test", min(rws):max(rws), 20) # -> Works
all.equal(rws, min(rws):max(rws)) # -> TRUE
identical(rws, min(rws):max(rws)) # -> FALSE
# Let's try with just a few rows (I typically work my way down using increasingly small row subsets to identify the offending one):
WB <- openxlsx2::wb_set_row_heights(WB, "test", c(99997, 99998, 99999), 20) # -> Works this time!
WB <- openxlsx2::wb_set_row_heights(WB, "test", 99997:99999, 20) # -> Works
# This only occurs if row 100000 is included in the range
WB <- openxlsx2::wb_set_row_heights(WB, "test", c(99999, 100000, 100001), 20) # -> Doesn't work
WB <- openxlsx2::wb_set_row_heights(WB, "test", c(99997, 99998, 99999) + 1, 20) # -> Also doesn't work
WB <- openxlsx2::wb_set_row_heights(WB, "test", 99999:100001, 20) # -> Works

Now, I just recently found out about ALTREPs, and whilst this is something I am absolutely not familiar with, this feels like it could be the cause. The 100000 feels like a boundary where R might change the underlying data representation, resulting in breaking the code here.

[JanMarvin]

Hi @Arthfael, thanks for the report. It’s a bit simpler, but nevertheless a bug.

as.character(100000)
#> "1e+05"

In our row table we search for "100000".

[Arthfael]

Ahhh of course! Anyway, I modified the calls in my code to the way which works in my example, so from my side the fix is not urgent ^^

…

On Tue, Oct 29, 2024 at 3:58 PM Jan Marvin Garbuszus < ***@***.***> wrote: Hi @Arthfael <https://github.com/Arthfael>, thanks for the report. It’s a bit simpler, but nevertheless a bug. as.character(100000)#> "1e+05" In our row table we search for "100000". — Reply to this email directly, view it on GitHub <#1167 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABSHV4JYTUQT6TUAOZVRFXLZ56PBNAVCNFSM6AAAAABQZXISMKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINBUGUYTKMJUHE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[JanMarvin]

I've added a PR in #1170. as.character(as.integer(rows)) should not break.