Merge branch 'master' into macroRemoval

Author: badasahog

DESCRIPTION

@@ -102,5 +102,6 @@ Authors@R: c(
   person("Aljaž", "Sluga",         role="ctb"),
   person("Bill", "Evans",          role="ctb"),
   person("Reino", "Bruner",        role="ctb"),
-  person(comment=c(github="@badasahog"), role="ctb")
+  person(comment=c(github="@badasahog"), role="ctb"),
+  person("Vinit", "Thakur",        role="ctb")
   )

NEWS.md

@@ -40,6 +40,8 @@
 
 10. On a heavily loaded machine, a `forder` thread could try to perform a zero-length copy from a null pointer, which was de-facto harmless but is against the C standard and was caught by additional CRAN checks, [#7051](https://github.com/Rdatatable/data.table/issues/7051). Thanks to @helske for the report and @aitap for the PR.
 
+11. Out of sample type bumps now respect `integer64=` selection, [#7032](https://github.com/Rdatatable/data.table/pull/7032).
+
 ### NOTES
 
 1. Continued work to remove non-API C functions, [#6180](https://github.com/Rdatatable/data.table/issues/6180). Thanks Ivan Krylov for the PRs and for writing a clear and concise guide about the R API: https://aitap.codeberg.page/R-api/.

README.md

@@ -19,12 +19,12 @@
 
 The `data.table` project uses a [custom governance agreement](./GOVERNANCE.md) 
 and is fiscally sponsored by [NumFOCUS](https://numfocus.org/). Consider making 
-a [tax-deductible donation](https://numfocus.org/donate-to-data-table) to help the project 
+a [tax-deductible donation](https://numfocus.org/project/data-table) to help the project 
 pay for developer time, professional services, travel, workshops, and a variety of other needs.
 
 <div align="center">
   <a href="https://numfocus.org/project/data-table">
-    <img height="60px" 
+    <img width="25%" 
          src="https://raw.githubusercontent.com/numfocus/templates/master/images/numfocus-logo.png" 
          align="center">
   </a>

inst/tests/tests.Rraw

@@ -3070,6 +3070,16 @@ if (test_bit64) test(1017.1, fread(f), copy(DT)[,A:=as.integer64(A)])
 test(1017.2, fread(f, integer64="character"), DT)
 unlink(f)
 
+DT = data.table(a=seq(10000), b="100")
+DT[111, b := "1000000000000"]
+f = tempfile()
+fwrite(DT, f)
+
+test(1017.3, fread(f, integer64="numeric"), fread(f, colClasses=c("integer", "numeric")))
+test(1017.4, fread(f, integer64="character"), fread(f, colClasses=c("integer", "character")))
+
+unlink(f)
+
 # ERANGE errno handled, #106 #4165
 test(1018.1, identical(fread("1.46761e-313\n"), data.table(V1=1.46761e-313)))
 test(1018.2, identical(fread("1.46761e+313\n"), data.table(V1=1.46761e+313)))

man/IDateTime.Rd

@@ -137,6 +137,11 @@ hours. Because \code{ITime} is stored in seconds, you can add it to a
 \code{POSIXct} object, but you should not add it to a \code{Date}
 object.
 
+For \code{as.ITime}, note that the string \code{"24:00:00"} is parsed as \code{"00:00:00"}.
+This is because the conversion uses \code{as.POSIXct}, which treats \code{"24:00:00"} as midnight of the next day. 
+This differs from ISO 8601 (which allows \code{"24:00:00"} to represent end-of-day), but aligns with POSIX standards.
+To represent end-of-day intervals, use \code{"23:59:59"} or arithmetic (e.g., \code{as.ITime("23:59:59") + 1L}).
+
 We also provide S3 methods to convert to and from \code{Date} and \code{POSIXct}.
 
 \code{ITime} is time zone-agnostic. When converting \code{ITime} and
@@ -245,6 +250,12 @@ identical(as.ITime("10:45"), methods::as("10:45", "ITime"))
 
 (t <- as.ITime("10:45:04", format = "\%H:\%M:\%S"))
 
+# "24:00:00" is parsed as "00:00:00"
+as.ITime("24:00:00")
+
+# Workaround for end-of-day: add 1 second to "23:59:59"
+as.ITime("23:59:59") + 1L
+
 as.POSIXct("2001-01-01") + as.ITime("10:45")
 
 datetime <- seq(as.POSIXct("2001-01-01"), as.POSIXct("2001-01-03"), by = "5 hour")

man/last.Rd

@@ -18,8 +18,8 @@ of \code{xts::first} is deployed. }
 \item{\dots}{ Not applicable for \code{data.table} first/last. Any arguments here
 are passed through to \code{xts}'s first/last. }
 }
-\details{
-Note: For zero-length vectors, \code{first(x)} and \code{last(x)} mimic \code{head(x, 1)} and \code{tail(x, 1)} by returning an empty vector instead of \code{NA}. However, unlike \code{head()}/\code{tail()} and base R subsetting (e.g., \code{x[1]}), they do not preserve attributes like names.
+\note{
+For zero-length vectors, \code{first(x)} and \code{last(x)} mimic \code{head(x, 1)} and \code{tail(x, 1)} by returning an empty vector instead of \code{NA}. However, unlike \code{head()}/\code{tail()} and base R subsetting (e.g., \code{x[1]}), they do not preserve attributes like names.
 }
 \value{
 If no other arguments are supplied it depends on the type of \code{x}. The first/last item

src/fread.c

@@ -2300,7 +2300,7 @@ int freadMain(freadMainArgs _args) {
   nth = imin(nJumps, nth);
 
   if (verbose) DTPRINT(_("[11] Read the data\n"));
-  while(true){  // we'll return here to reread any columns with out-of-sample type exceptions, or dirty jumps
+  for (;;) {  // we'll return here to reread any columns with out-of-sample type exceptions, or dirty jumps
     restartTeam = false;
     if (verbose)
       DTPRINT("  jumps=[%d..%d), chunk_size=%zu, total_size=%td\n", jump0, nJumps, chunkBytes, eof-pos); // # notranslate
@@ -2510,7 +2510,12 @@ int freadMain(freadMainArgs _args) {
               // check that the new type is sufficient for the rest of the column (and any other columns also in out-of-sample bump status) to be
               // sure a single re-read will definitely work.
               while (++absType<CT_STRING && disabled_parsers[absType]) {};
-              thisType = TOGGLE_BUMP(absType);
+
+              if(args.readInt64As != CT_INT64 && absType == CT_INT64)
+                thisType = TOGGLE_BUMP(args.readInt64As);
+              else
+                thisType = TOGGLE_BUMP(absType);
+
               tch = fieldStart;
             }
 

src/fread.h

@@ -167,6 +167,9 @@ typedef struct freadMainArgs
   // should datetime with no Z or UTZ-offset be read as UTC?
   bool noTZasUTC;
 
+  // Integer64 remap
+  colType readInt64As;
+
   char _padding[1];
 
   // Any additional implementation-specific parameters.

src/freadR.c

@@ -177,6 +177,8 @@ SEXP freadR(
     readInt64As = CT_FLOAT64;
   } else STOP(_("Invalid value integer64='%s'. Must be 'integer64', 'character', 'double' or 'numeric'"), tt);
 
+  args.readInt64As = readInt64As;
+
   colClassesSxp = colClassesArg;
 
   selectSxp = selectArg;