On Thu, Jul 22, 2021 at 04:56:56PM +0200, Peter Eisentraut wrote:
> It seems to me that when using the pg_amcheck --startblock and --endblock
> options on platforms where sizeof(long) == 4, you cannot specify higher
> block numbers (unless you do tricks with negative numbers). The attached
> patch should fix this by using strtoul() instead of strtol(). I also
> tightened up the number scanning a bit in other ways, similar to the code in
> other frontend utilities. I know some people have been working on
> tightening all this up. Please check that it's up to speed.
Yeah, some work is happening to tighten all that. Saying that, the
first round of review I did for the option parsing is not involving
option types other than int32, so the block options of pg_amcheck are
not changing for now, and what you are suggesting here is fine for the
moment for 14~. Still, note that I am planning to change that on HEAD
with an option parsing API for int64 that could be used for block
numbers, eliminating for example the 4 translatable strings for the
code being changed here.
--
Michael
