Superfluous #ifdef LV_FS_WIN32_PATH in lv_fs_win32.c?

[vwheeler63]

Introduce the problem

In having a look at issue #7601 a few minutes ago, I (once again) looked into how the LV_FS_WIN32_PATH macro is used, and I noticed an inconsistency in how it is used in lv_fs_win32.c. On line 363 in the fs_dir_open() function:

#ifdef LV_FS_WIN32_PATH
    lv_snprintf(buf, sizeof(buf), LV_FS_WIN32_PATH "%s\\*", path);
#else
    lv_snprintf(buf, sizeof(buf), "%s\\*", path);
#endif

but on line 226 in fs_open():

    lv_snprintf(buf, sizeof(buf), LV_FS_WIN32_PATH "%s", path);

and there is no protective #ifdef.

At first, I thought the fault was on line 226, and to prove it, I undefined LV_FS_WIN32_PATH in lv_conf.h and tried to compile lv_fs_win32.c. To my surprise, it compiled without errors. And then I realized: lv_conf_internal.h supplies it with a valid default value of an empty string "". That it is not possible for an end user (programmer) to accidentally not define LV_FS_WIN32_PATH.

So it now appears to me that the first use of it on line 363 is in fact "the inconsistency" -- that no alternate compilation directive should be used. Reason this is important: because it can mis-direct developers using LVGL to thinking that LV_FS_WIN32_PATH can somehow not be defined when that code is compiled. (Note that the entire file is protected with #if LV_USE_FS_WIN32 != '\0'.)

As a side note (different topic): #if LV_USE_FS_WIN32 != '\0' should probably be #if LV_USE_FS_WIN32 != 0. While I am aware that \0 also resolved to 0, there is no reason it should be represented as a char data type.

Proposal

Line 363 be changed to:

    lv_snprintf(buf, sizeof(buf), LV_FS_WIN32_PATH "%s\\*", path);

[kisvegabor]

I agree with both!

#if LV_USE_FS_WIN32 != '\0'

Probably is was confuse with LV_FS_WIN32_LETTER where we use \0 to indicate that the value is a char.