Hi,
> -----Original Message-----
> From: Johannes Schindelin [mailto:[EMAIL PROTECTED]
> Sent: Sunday, February 03, 2008 6:58 PM
> On Sat, 2 Feb 2008, Kirill wrote:
> > TODO: decide on a structure and actions of added menu items.
> Yep. I think a good starting point is to look at what
> TortoiseCVS/TortoiseSVN provide.
I thought you were not in favour of that because git has a very powerful
feature - index. Did I miss something?
> Personally, the thing I used most (back when I still had to work with
> CVS on Windows), was the diff operation.
> Another really nice thing was to see the history of a file (I guess
> this would be "gitk HEAD -- <file>").
Totally agree on both points!
> > @@ -147,13 +224,11 @@ static STDMETHODIMP invoke_command(void *p,
> > STARTUPINFO si = { sizeof(si) };
> > PROCESS_INFORMATION pi;
> >
> > - TCHAR command[1024];
> > + char command[1024];
> > const char *wd;
> > DWORD dwAttr, fa;
> >
> > - wsprintf(command, TEXT("wish.exe \"%s/bin/git-gui\""),
> > - msys_path());
> > -
> > + sprintf(command, "wish.exe \"%s/bin/git-gui\"",
> > + msys_path());
> >
>
> This line should be empty, but has white space.
I'm not sure, what you are referring to? But nevertheless, I'll keep empty
lines empty.
> Besides, this whole hunk is unrelated to the commit message, correct?
>
> > @@ -202,7 +277,7 @@ static STDMETHODIMP get_command_string(void *p,
> UINT id,
> >
> >
> > if (flags & GCS_HELPTEXT) {
> > - LPCTSTR text = _T("Launch the GIT Gui in the local or
> chosen directory.");
> > + char *text = "Launch the GIT Gui in the local or chosen
> directory.";
>
> Same goes for this hunk...
Oops! Sorry! These two changes are asking for a completely separate
discussion. Feel free to take it to another thread, but because they showed
up...
I'm not quite sure how it's handled on Unix or in git.git for this matter,
but the TCHAR existence is somewhat artificial. IIRC, Microsoft introduced
TCHAR (and _T() macro) in their environment quite some time ago to simplify
a migration from MBCS to Unicode. Basically, the idea is to have a type that
is defined differently #ifdef UNICODE. Now, it seems like in Cheetah, TCHAR
sometimes is used as a synonym to wchar_t, which it is not, unless I
overlooked something obvious in gcc behaviour. And sometimes it just seems
like a copy-and-paste from a different universe.
Hence, under the assumption that gcc does not auto-magically defines UNICODE
and TCHAR is indeed MS-specific:
- I have no idea how the first hunk (with wsprintf(char *, char *, char *);)
ever worked;
- as of the second hunk (with LPCTSTR), given that Cheetah is mostly written
to be MBCS, I don't see a point to use MS-specific, transitional version of
const char * (that's what LPCTSTR boils down to) in only one-or-two places.
So, a big question of the day: do we write Cheetah as MBCS? or as Unicode?
or in transitional fashion, using TCHAR & Co.?
> Okay, reading my comments again, it seems like all of them are
> negative. Please bear in mind, though...
Yeah! Without this part, I was almost ready to declare the week completely
ruined (nobody likes my code :) Seriously:
- I greatly appreciate all the efforts!
- I greatly appreciate style comments! I guess I'll need to find out what's
wrong with C++-style comments, and overcome an urge to put couple "extra"
spaces... But bear with me, would you? Please?
- a quote from someone of the great "I am a lazy bastard..." is applicable
to me too.
Oh, well... the last one was not totally serious, but... like a local
comercial for Wal-Mart says "It's tru-u-ue"
Thanks!
--
Kirill Alexandrov.
