ConnectBlock(): fix error() format to be unsigned #2022

• I introduced the wrong format macro with my former patch (#2018), this needs to be signed not unsigned (thanks Luke-Jr)

These were the warning my patch fixes: src\main.cpp: In member function 'bool CBlock::ConnectBlock(CBlockIndex*, CCoinsViewCache&, bool)': src\main.cpp:warning: unknown conversion type character 'l' in format [-Wformat] src\main.cpp:warning: unknown conversion type character 'l' in format [-Wformat] src\main.cpp:warning: too many arguments for format [-Wformat-extra-args]

Looks like a bug in Microsoft printf family of functions :(

ACK for 0.7.2 & master

Thanks for taking a look and I wanted to clarify why I patched it :).

ACK

Please fix the commit message. The one-line summary of your change, as it appears in "git shortlog" and other tools, is wholly useless: "fix pull #2018"

That is utterly meaningless without further research; it tells nothing about what code is modified, or how it was modified, or why it needed modification.

We use "git shortlog" to generate summaries for release notes and other purposes. The first line is a summary of the change, and for this, probably should be something like "ConnectBlock: fix error() format to be unsigned" etc.

How is this a bug in the Microsoft printf function @Luke-Jr?

@laanwj %lld is standard C printf format, but not supported by Microsoft's printf: http://sourceforge.net/tracker/?func=detail&atid=102435&aid=1963136&group_id=2435

Ooh, okay something made by MS is incompatible to the standards, now that's a now one, let's call the press :)

I still don't get those warnings though. We define:

#ifndef PRI64d
#if defined(_MSC_VER) || defined(__MSVCRT__)
#define PRI64d  "I64d"
#define PRI64u  "I64u"
#define PRI64x  "I64x"
#else
#define PRI64d  "lld"
#define PRI64u  "llu"
#define PRI64x  "llx"
#endif
#endif

Which means that in the case of MS, we use "special" retarded character sequences. The thing is, there is no 'l' in any of those. So where does that warning come from?

@laanwj The original code which Diapolo is referring to used %lld; he replaced that with PRI64u, but it needs to be PRI64d since the argument is signed.

Right, I get it now, thanks

@jgarzik Sorry, yes it was a weak commit message :), fixed it!