If %e specifiers are enabled without enabling %f specifiers, then neither are provided.

[mickjc750]

Enabling option PRINTF_SUPPORT_EXPONENTIAL_SPECIFIERS but not PRINTF_SUPPORT_FLOAT_SPECIFIERS.
Yields neither..
This may occur if an application really only wants the exponent format.

[eyalroz]

I don't actually see this happening...

/home/eyalroz/src/mine/mpaland-printf/test/test_suite.cpp:849: PASSED:
  CHECK( !strcmp(buffer, "4.895512e+04") )
with expansion:
  true
with messages:
  test::sprintf_ arguments, replicated ( "arg := arg" ):
  ----
  "%e" := "%e"
  48955.125 := 48955.125
  ----
  Resulting buffer contents: "4.895512e+04"

etc.

[eyalroz]

@mickjc750 : ping.

[mickjc750]

@eyalroz
I'm not sure how to reproduce your snippet from the test_suit. I originally saw this issue when compiling the .c and .h files alongside the AVR test. But I can reproduce it on a desktop with the following steps :

Edit the printf_config.h.in to the following:
#cmakedefine PRINTF_SUPPORT_DECIMAL_SPECIFIERS 0
#cmakedefine PRINTF_SUPPORT_EXPONENTIAL_SPECIFIERS 1
#cmakedefine PRINTF_SUPPORT_LONG_LONG 1
#cmakedefine PRINTF_ALIAS_STANDARD_FUNCTION_NAMES 1

Edit the autotest.cpp to include the following on line 118 after the usage message:
sprintf_(txt, "%%e = %e %%f = %f", 8E41, 0.97341);
printf("%s\n",txt);

Then:
cmake .
make
sudo make install
cd test
cmake .
cp ../include/printf_config.h .
make
./autotest

The output after the usage message shows:
%e = e %f = f

[eyalroz]

Well, it seems I have a bug with the CMake generation of printf_config.h. When I fix that (see #41), I see what you're seeing.

[eyalroz]

@mickjc750 : Please try this out.

[mickjc750]

@eyalroz confirmed fixed.