Add error and range-checking parsers for unsigned 32 and 64 bit numbers.
The 32-bit variant is required for parsing sequence numbers from the command line in bitcoin-tx (see #8164 for discussion). I've thrown in the 64-bit variant as a bonus, as I'm sure it will be needed at some
point.
Also adds tests, and updates developer-notes.md.
Add error and range-checking parsers for unsigned 32 and 64 bit numbers.
The 32-bit variant is required for parsing sequence numbers from the
command line in `bitcoin-tx` (see #8164 for discussion). I've thrown in
the 64-bit variant as a bonus, as I'm sure it will be needed at some
point.
Also adds tests, and updates `developer-notes.md`.
utACK e012f3c. I think we also need them in init.cpp
utACK e012f3cea0ca4096dd4dd59a356a973c43651912
472 | + unsigned long int n = strtoul(str.c_str(), &endp, 10); 473 | + if(out) *out = (uint32_t)n; 474 | + // Note that strtoul returns a *unsigned long int*, so even if it doesn't report a over/underflow 475 | + // we still have to check that the returned value is within the range of an *uint32_t*. On 64-bit 476 | + // platforms the size of these types may be different. 477 | + return endp && *endp == 0 && !errno &&
Instead of checking whether *endp == 0, shouldn't you check whether endp == str.c_str() + str.size() (otherwise you'd accept strings that have a 0 inside)
Ah, ParsePrechecks catches this.
Yes, that solution would be somewhat more elegant. Indeed, ParsePrechecks checks some number-parsing preconditions such as that one.
I'm thinking about (at some point) implementing these functions from scratch instead of calling out to the C API. It wouldn't even be much longer, as there are just too many exceptions. This still doesn't exclude strto*l accepting locale-specific separators, for example. Not a big deal for current uses, but I hate such surprises.
In any case introducing them here (with tests) allows swapping out the implementation later.
Yes, I was about to suggest that a from scratch implementation might be easier than trying to put a harness around existing libc calls. But let's not do that now. It works and seems sufficiently tested.
utACK e012f3cea0ca4096dd4dd59a356a973c43651912
