Miscelleanous fixes #17
Conversation
Accompanies neutrinolabs/xrdp#467
| @@ -9,7 +9,7 @@ AC_PROG_CC | |||
| AC_C_CONST | |||
| AC_PROG_LIBTOOL | |||
|
|
|||
| AM_CONDITIONAL(GOT_PREFIX, test "x${prefix}" != "xNONE"]) | |||
| AM_CONDITIONAL(GOT_PREFIX, [test "x${prefix}" != "xNONE" && test "x${prefix}" != "x/usr"]) | |||
There was a problem hiding this comment.
GOT_PREFIX is not needed at all. My pull request #13 removes it. The problem with GOT_PREFIX is that it doesn't have a clear semantic - when it should be defined and what it should do. The implication is that if the user installs software to some location, includes and libraries already in that location should be used. The problem is that it's not always the case, and it's easier to add a directory to the includes than to remove it.
| @@ -668,7 +670,8 @@ rdpkeybControl(DeviceIntPtr device, int what) | |||
| switch (what) | |||
| { | |||
| case DEVICE_INIT: | |||
| rdpkeybDeviceInit(device, &keySyms, modMap); | |||
| if (rdpkeybDeviceInit(device, &keySyms, modMap)) | |||
| return BadAlloc; | |||
There was a problem hiding this comment.
I don't like returning BadAlloc here. It created a tight coupling. One function knows why another function fails. Perhaps rdpkeybDeviceInit() itself could return BadAlloc, and its callers could pass it further up.
|
The PR can be closed now. GOT_PREFIX has already been removed. Delaying exit on memory allocation failure is wrong in my opinion, and we haven't heard from the PR author. |
Accompanies neutrinolabs/xrdp#467