Re: Yet another look at the apr_off_t problem

Ben Reser wrote:
> On Fri, Jan 23, 2004 at 05:44:21PM -0800, Ben Reser wrote:
>
>>Make the blame fix because the line_no parameter doesn't make any sense
>>being an apr_off_t type. We change that to apr_uint64_t. We put that
>>in 0.37.0. This fixes the blame problem.
>
> The blame patch is now attached to Issue #1705 as attachment 291.

I understand the explanation that has been given for why the current "apr_off_t" is unsuitable for use in a library API, but I am not completely familiar with library API/ABI design issues. The use of "apr_uint64_t" to hold a line number seems to me very ugly. A line number is not logically a 64-bit type, it is an unbounded type. Interfaces within an application are normally best written with plain "int". If "int" is not big enough for someone's line numbers, then they almost certainly need a bigger computer (one which is natively 64-bit or 128-bit).

Is there an ABI-related reason not to use plain "int" in library APIs? If so, what about our several other APIs that already use plain "int"?

- Julian

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Jan 24 16:33:30 2004