On Tue, 24 Jul 2018 at 13:49, Robert Elz <k...@munnari.oz.au> wrote: > I have just added support for the F a and A formats to printf(1). > > These are not required by POSIX, but nor are the e E f g or G formats, > which we did support. and supporting those 5 without the other 3 makes > allmost no sense (in the code they are essentially free, the only real cost > is a little more text in the man page). > > Those are done. > > But I also have support for the "data width modifiers" %ld %jx %Lf ... > (all of them I believe, as supported by printf(3), though not all are fully > supported., there is no %lc or %ls for example - those just generate > errors.) > > These are also not required by POSIX. They also mostly do nothing at > all (do not in any way change the output) - with a couple of exceptions. > > Just like printf(3) %h truncates the argument to short (or unsigned short, > as appropriate) and %hh truncates to char (or unsigned char). And %Lf > does (and all the other float formats with L do) use long double, rather > than > double. > > The point is just that it makes it easier to use printf(1) to test a > proposed > format string to be used with printf(3), and make sure that the appearance > of the output is what you want, without needing to either recompile the > program all the time, or write a one-off special program that takes a > format > string and data to print, and prints it (which is, really, what printf(1) > is). > > At least 'l' (as in %ld) is supported in some other printf(1) > imlementations. > > Any objections to adding support for these "data width" modifiers ? > > kre > > No objections, I'd love to see these. I have trouble remembering printf field specifiers, and almost always have to resort to a trial and error compile-run-wtf frenzy when I least need it.
Thanks, Alistair
