On Tue, Apr 21, 2009 at 2:51 PM, Nicolas Le Cam <niko.le...@gmail.com> wrote:
> 2009/4/21 James Hawkins <trui...@gmail.com>:
>> On Tue, Apr 21, 2009 at 2:30 PM, Nicolas Le Cam <niko.le...@gmail.com> wrote:
>>> This one finally fixes current relative path test to expect correct
>>> value.
>>>
>>
>> + drives = GetLogicalDrives();
>> + lstrcpyA(path, "A:\\");
>> + for (i = 0; i < 26; path[0] = '\0', i++)
>> + {
>> + if (!(drives & (1 << i)))
>> + continue;
>> +
>> + path[0] = 'A' + i;
>> + if (GetDriveType(path) != DRIVE_FIXED)
>> + continue;
>> +
>> + lstrcatA(path + 3, CURR_DIR + 3);
>> + attr = GetFileAttributesA(path);
>> + if (attr != INVALID_FILE_ATTRIBUTES && (attr &
>> FILE_ATTRIBUTE_DIRECTORY))
>> + {
>> + if (path[lstrlenA(path)-1] != '\\')
>> + lstrcatA(path, "\\");
>> + break;
>> + }
>> + path[3] = '\0';
>> + }
>>
>>
>> Same as 2/4. Please refactor and add many nice comments detailing
>> exactly what is happening in native msi to make this necessary.
>>
>> --
>> James Hawkins
>>
>
> Hi James, thanks for review.
> Will resubmit tomorrow.
>
> To answer your question, basically native msi is doing exactly what
> wine is doing (ACTION_SearchDirectory in appsearch.c) except for the
> corner case I'm trying to solve, too bad I did find this piece of code
> after hours of testing to found out what is native behavior, at least
> it has confirmed what I have found :).
>
What I mean is you need a comment detailing the preconditions, the
logic used in msi, and the end result. Someone with a basic
understanding of the AppSearch action should be able to read the test
and understand what is happening.
--
James Hawkins
