[lldb-vscod] fix build with NDEBUG on windows

[SquallATF]

Summary

_setmode in assert will not run when build with NDEBUG

[mstorsjo]

Good catch, thanks! But would we need to add something like (void)result; to silence potential compiler warnings about an unused (or set but not read) variable?

[labath]

LGTM, modulo the (void) result stuff. Do you need someone to commit this for you?

[SquallATF]

In D69612#1726829, @labath wrote:

LGTM, modulo the (void) result stuff. Do you need someone to commit this for you?

Yes I need.

[mstorsjo]

I presume you applied the patch with arc patch? It seems to apply the Phabricator username as git user, which is a bit odd, as it is customary to use realnames in the git author field. Earlier, when everything ended up mangled by svn, everything ended up normalized to the committer's name anyway, but as git preserves the author names, it would probably be good if the arc workflow could use the realname instead.

[labath]

Yes, I used "arc patch". I did check that it preserved the authorship information, but I did not go into the details of what's in the author field TBH. I agree it would be better to have the full name there, but I don't know if there's a way to change that (short of people doing "git commit --amend --author=XXX" manually, which everyone is going to forget).

[amccarth]

In D69612#1726849, @SquallATF wrote:

In D69612#1726829, @labath wrote:

LGTM, modulo the (void) result stuff. Do you need someone to commit this for you?

Yes I need.

Was the (void) result fix applied before this was committed?

[labath]

It ended up being a separate commit (2d1a0dfe) because I don't know how to use git. :P As far as change attribution is concerned, that may be even for the better. :)

[labath]

In D69612#1726968, @labath wrote:

Yes, I used "arc patch". I did check that it preserved the authorship information, but I did not go into the details of what's in the author field TBH. I agree it would be better to have the full name there, but I don't know if there's a way to change that (short of people doing "git commit --amend --author=XXX" manually, which everyone is going to forget).

BTW, I've just done another "arc patch" and the author information came out right this time. So it looks like arc is able to fetch/set the git author correctly at least in some cases.

lldb/tools/lldb-vscode/VSCode.cpp

: launch_info(nullptr), variables(), broadcaster("lldb-vscode"),
            {"swift_throw", "Swift Throw", lldb::eLanguageTypeSwift}}),
       focus_tid(LLDB_INVALID_THREAD_ID), sent_terminated_event(false),
       stop_at_entry(false) {
   const char *log_file_path = getenv("LLDBVSCODE_LOG");
 #if defined(_WIN32)
 // Windows opens stdout and stdin in text mode which converts \n to 13,10
 // while the value is just 10 on Darwin/Linux. Setting the file mode to binary
 // fixes this.
-  assert(_setmode(fileno(stdout), _O_BINARY));
-  assert(_setmode(fileno(stdin), _O_BINARY));
+  int result = _setmode(fileno(stdout), _O_BINARY);
+  assert(result);
+  result = _setmode(fileno(stdin), _O_BINARY);
+  assert(result);
 #endif
   if (log_file_path)
     log.reset(new std::ofstream(log_file_path));
 }
 
 VSCode::~VSCode() {
 }
 
void VSCode::RunStopCommands() {
   RunLLDBCommands("Running stopCommands:", stop_commands);
 }
 
 void VSCode::RunExitCommands() {
   RunLLDBCommands("Running exitCommands:", exit_commands);
 }
 
 } // namespace lldb_vscode
-