On Aug 22, 7:55 pm, [EMAIL PROTECTED] wrote:
> [scriptable, uuid(096B7F86-2E47-4BF1-9A18-0D46D81EAEAC)]
looks like an MSCOM interface name, moderately confusing but ok:
> interface IDNA : nsISupports
> {
attributes and methods should be interCaps not IniitalCaps:
> readonly attribute string Id;
> string GetLastError();
You shouldn't use the "string" type, doing so will get you into the
trouble you're about to get into.
> #define DNA_CID {0x096b7f86, 0x2e47, 0x4bf1, { 0x9a, 0x18, 0x0d, 0x46,
> 0xd8, 0x1e, 0xae, 0xac }}
Class IDs and Interface IDs are in different namespaces, but in order
to avoid people being confused and think they relate, let me tell you
that you should *NOT* share GUIDs between Interfaces and CIDs.
mozilla style would be char mErrorMessage[200] or something (better
nsCString mError;)
> char m_errormsg[200];
proper signature, yay!
> NS_IMETHODIMP DNA::GetId(char * *aId)
> {
> char* ret = NULL;
strange, but OK.
worrisome:
> ret = reg.GetRegistryId();
>
> if( ret )
> {
need to verify GetRegistryId doesn't misbehave.
> *aId = ret;
> strncpy( m_errormsg, reg.GetLastError(), sizeof( m_errormsg )
> );
I wouldn't do this:
> *aId = "";
Your C++ compiler should complain as you're assigning a const char[]
to a char*. And if it manages to work, you manage to mix memory
allocators which should cause a crash. (Hopefully sooner, but probably
later.)
correct signature for XPCOM if you use it correctly.
> NS_IMETHODIMP DNA::GetLastError(char **_retval)
> {
Which you don't:
> *_retval = m_errormsg;
This is wrong. You've used the wrong memory allocator (C++ object heap
for DNA instead of XPCOM nsMemory/NS_Alloc).
FWIW, gecko manages to mess up occasionally w/ this stuff, but your
versions of the mistakes are worse and will crash sooner than ours.
_______________________________________________
dev-tech-xpcom mailing list
dev-tech-xpcom@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-tech-xpcom
