Fix char/wchar/tchar inconsistencies in PythonService.cpp

[joankaradimov]

While trying to compile the code with gcc I got a bunch of errors. I fix them in this PR. Most of the fixes make sense, but there is one dodgy part that needs to be reviewed:

#ifdef UNICODE
MultiByteToWideChar(CP_ACP, 0, szmbTraceback, tb_len, szTraceback, tb_len);
#else
strncpy(szTraceback, szmbTraceback, tb_len);
#endif

[mhammond]

Fixing inconsistencies sounds great, but if possible, I'd rather consistently use WCHAR - there's no reason to have this building with narrow characters. setup.py tries to force unicode, but does this cause an issue with gcc?

[joankaradimov]

setup.py tries to force unicode, but does this cause an issue with gcc?

No, gcc/mingw handles UNICODE/wchar_t just fine.

Fixing inconsistencies sounds great, but if possible, I'd rather consistently use WCHAR

I could do that too. The problem is this function:

static BOOL ReportError(DWORD, LPCTSTR *inserts = NULL, WORD errorType = EVENTLOG_ERROR_TYPE);

It uses LPCTSTR for the inserts argument and not LPCWSTR. And the function is used quite a bit. I just went with what involves less changes. But I could make the inserts argument be wide string and change all of its usages.

[mhammond]

It uses LPCTSTR for the inserts argument and not LPCWSTR

But aren't they always identical in this case because this is never built without UNICODE defined?

[joankaradimov]

I didn't say it's an error, I said it's an inconsistency :)

[mhammond]

As mentioned above, I'd really prefer this consistently use WCHAR instead of TCHAR.

[joankaradimov]

I'll try to figure out if I can improve the current state of things and force push new commits. If I can't find a way I'll just close the PR.

[joankaradimov]

I'm closing this. Nothing I do is an improvement on either the current code or the code in the PR.