MCSTR() breaks providers lookups

[CodaFi]

While we're at it, we should also synthesize the identifier property

[dinhvh]

It's not clear what this issue is about. Could you clarify?

[CodaFi]

Set a breakpoint in fillWithInfo() and you can see that all of the objectForKey() calls are failing because MCSTR() does "" "str" "" instead of "str"

[dinhvh]

When you're talking about code, it's more clear if you reference which file to look into. fillWithInfo() name is used at several locations.

Regarding the issue itself:

I don't think that's the reason of your issue. Keep investigating about your specific issue.
If MCSTR() was really broken, it would fail everywhere.

MCSTR() does "" str "", which concatenate an empty string with str with an empty string.
It makes sure that str is a constant.

[dinhvh]

Reopen if that's really the issue and give more details.

[CodaFi]

But it fails specifically for the providers dictionary because the JSON keys aren't made with MCSTR(), they're just parsed out into it. It works everywhere else because everywhere else uses MCSTR().

Here, these lines fail for every provider HashMap given to them:

https://github.com/MailCore/mailcore2/blob/master/src/core/provider/MCMailProvider.cc#L62
https://github.com/MailCore/mailcore2/blob/master/src/core/provider/MCMailProvider.cc#L66

And because they fail, every one of the folder lookup methods afterwards dereferences a NULL pointer and causes a fatal crash.

[CodaFi]

If I replace those MCSTR()'s with mailcore::String::stringWithUTF8Format() it works like a charm.

[dinhvh]

Therefore, that's an issue with String::uniquedStringWithUTF8Characters().
Do you have a minimal code that reproduces the issue?

Could you try to apply that patch?

diff --git a/src/core/basetypes/MCString.cc b/src/core/basetypes/MCString.cc
index 6710e59..2398958 100644
--- a/src/core/basetypes/MCString.cc
+++ b/src/core/basetypes/MCString.cc
@@ -1975,6 +1975,7 @@ String * String::substringWithRange(Range range)
}

static chash * uniquedStringHash = NULL;
+static pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;

static void initUniquedStringHash()
{
@@ -1991,14 +1992,17 @@ String * String::uniquedStringWithUTF8Characters(const char * UTF8Characters)
pthread_once(&once, initUniquedStringHash);
key.data = (void *) UTF8Characters;
key.len = (unsigned int) strlen(UTF8Characters);

• pthread_mutex_lock(&lock);
  r = chash_get(uniquedStringHash, &key, &value);
  if (r == 0) {

•    pthread_mutex_unlock(&lock);
   return (String *) value.data;
  

  }
  else {
  value.data = new String(UTF8Characters);
  value.len = 0;
  chash_set(uniquedStringHash, &key, &value, NULL);

•    pthread_mutex_unlock(&lock);
   return (String *) value.data;
  

  }
  }

[dinhvh]

Patch not broken by formatting is here:
https://gist.github.com/dinhviethoa/5692527
Let me know if it solves your issue.

[CodaFi]

Yep, that patch solves it. Sorry I couldn't answer you earlier, I was integrating the framework and had to fix a ton of warnings before I could test anything.

[dinhvh]

Fixed with b88d4be