Closed
Bug 68790
Opened 24 years ago
Closed 24 years ago
fixing C++ standard problems in some headers
Categories
(SeaMonkey :: Build Config, defect)
SeaMonkey
Build Config
Tracking
(Not tracked)
VERIFIED
FIXED
mozilla0.9
People
(Reporter: brendan, Assigned: harishd)
Details
Attachments
(1 file)
13.18 KB,
patch
|
Details | Diff | Splinter Review |
I'm slowly trying to massage mozilla so it will build successfully with the
current development copy of gcc, which is destined to come out as gcc 3.0.
I had to make some changes to headers to not introduce declarations of ostream
and other things that were plain classes, but are now templates according to the
C++ standard.
In htmlparser/src/nsHTMLContentSinkStream.h, I had to make it also see the newer
g++:
#if !defined(XP_MAC) && !defined(__KCC) && !(defined(__GNUC__) &&
defined(__GNUC_MINOR__) && __GNUC_MINOR__ >= 97)
class ostream;
#endif
B
Comment 1•24 years ago
|
||
Marking NEW to get someone to get it into tbe build.
This is akkana's realm. Handing over to her.
Assignee: harishd → akkana
Comment 4•24 years ago
|
||
Why are we even building nsHTMLContentSinkStream? I thought that was replaced
by the html content serializer. Johnny and Joe, what's up with this? Is this
all duplicate code with the html serializer now?
Assignee: akkana → jst
Comment 5•24 years ago
|
||
after staring at lxr a while I think no one is using it. But it is a supported
component in parser, ie somone in the future could ask the parser "hey give me an
nsHTMLContentSinkStream". right now no one is I think (though several files
include it's header and a couple even have the cid defined).
It's worth noting that this thing isn't really the same as the serializer. The
serializer is for going from content to stream. nsHTMLContentSinkStream is just
a sink for the parser that builds a stream rather than content. AFAIK it's
really for parser testing. Seems like a useful thing. I wouldn't nuke it,
though we could take it out of the opt build.
Comment 6•24 years ago
|
||
The html serializer can act as a content sink, too, or that was my understanding.
The code in nsHTMLContentSinkStream is orphaned and will bitrot, so it should be
removed and anyone who wants a content sink that does that should ask for the
serializer instead. (Johnny can correct me if I'm wrong about that.) Having
two classes which do the same thing, where only one of them is used in our code
but a client can ask for the other one, is just asking for trouble.
Comment 7•24 years ago
|
||
ok, i'll shut up. Looking at the serializers i see some methods that use parser
nodes and some that use content, so i guess they can be parser sinks. In which
casde I agree that the other sink should go.
Comment 8•24 years ago
|
||
The sink is currently needed in one place only, and that's in
nsParser::IsValidFragment(), which is called from nsRange::IsValidFragmen() that
is defined in nsIDOMNSRange, i.e. in our proprietary range interface.
nsIDOMNSRange::IsValidFragment() is not used anywhere in our codebase, and I
talked to Vidur and he said if it's not used and nobody is planning on using it
any time soon, let's get rid of it. If we'd do that we can remove the
HTMLContentSinkStream once n' for all. Editor people, thoughts?
Comment 9•24 years ago
|
||
If it's no longer needed for CreateContextualFragment (and if Vidur says it
isn't, then it isn't), let's nuke that puppy.
Comment 10•24 years ago
|
||
I'm sure Harish would love to rip this code out of the parser :-) Reassigning.
Assignee: jst → harishd
OS: Linux → All
Hardware: PC → All
Assignee | ||
Comment 11•24 years ago
|
||
Comment 12•24 years ago
|
||
sr=jst, when you check in, please remove the nsHTMLContentSinkStream_CID
definition from xpcom/base/IIDS.h too.
Like Johnny said, remove also nsHTMLContentSinkStream_CID in xpcom/base/IIDS.h.
Also, remove nsHTMLContentSinkStream.h export in htmlparser/src/MANIFEST.
With those fixes, r=heikki.
Assignee | ||
Comment 14•24 years ago
|
||
Fix is in... Marking FIXED.
Status: ASSIGNED → RESOLVED
Closed: 24 years ago
Resolution: --- → FIXED
Updated•20 years ago
|
Product: Browser → Seamonkey
You need to log in
before you can comment on or make changes to this bug.
Description
•