Return to Answer

printf("%s", (void*) NULL) is UB, so doesn't need testing. Thanks @chux.

Source Link

edited Jul 5, 2018 at 7:05

Some small items, additional to the great review by Chux.

You don't seem to use anything from <assert.h> or <float.h>, so no need to include those.
The missing features are not documented anywhere, so we have to guess what subset of printf() is supported.
Test cases are missing for ±infinity±∞ and not-a-number values with the %f specifier and for null pointers with the %s specifier.
Precision specifiers don't work properly with %s:
```
  myprintf("%.5s", "abcdefg");  // crashes
```
Unnecessary obfuscation: *(s_p+1) is much more recognisable when written as s_p[1].
The fallthrough from case '.' to default causes the second conversion specifier to be printed literally in this example (and could be even more serious if the second conversion were %%):
```
  myprintf("%.1f%f\n", 1.0, 0.0);
```

Some small items, additional to the great review by Chux.

You don't seem to use anything from <assert.h> or <float.h>, so no need to include those.
The missing features are not documented anywhere, so we have to guess what subset of printf() is supported.
Test cases are missing for ±infinity and not-a-number values with the %f specifier and for null pointers with the %s specifier.
Precision specifiers don't work properly with %s:
```
  myprintf("%.5s", "abcdefg");  // crashes
```
Unnecessary obfuscation: *(s_p+1) is much more recognisable when written as s_p[1].
The fallthrough from case '.' to default causes the second conversion specifier to be printed literally in this example (and could be even more serious if the second conversion were %%):
```
  myprintf("%.1f%f\n", 1.0, 0.0);
```

Some small items, additional to the great review by Chux.

You don't seem to use anything from <assert.h> or <float.h>, so no need to include those.
The missing features are not documented anywhere, so we have to guess what subset of printf() is supported.
Test cases are missing for ±∞ and not-a-number values with the %f specifier.
Precision specifiers don't work properly with %s:
```
  myprintf("%.5s", "abcdefg");  // crashes
```
Unnecessary obfuscation: *(s_p+1) is much more recognisable when written as s_p[1].
The fallthrough from case '.' to default causes the second conversion specifier to be printed literally in this example (and could be even more serious if the second conversion were %%):
```
  myprintf("%.1f%f\n", 1.0, 0.0);
```

Source Link

created Jul 4, 2018 at 17:42

Some small items, additional to the great review by Chux.

You don't seem to use anything from <assert.h> or <float.h>, so no need to include those.
The missing features are not documented anywhere, so we have to guess what subset of printf() is supported.
Test cases are missing for ±infinity and not-a-number values with the %f specifier and for null pointers with the %s specifier.
Precision specifiers don't work properly with %s:
```
  myprintf("%.5s", "abcdefg");  // crashes
```
Unnecessary obfuscation: *(s_p+1) is much more recognisable when written as s_p[1].
The fallthrough from case '.' to default causes the second conversion specifier to be printed literally in this example (and could be even more serious if the second conversion were %%):
```
  myprintf("%.1f%f\n", 1.0, 0.0);
```