CVSROOT: /sources/gnash
Module name: gnash
Changes by: Sandro Santilli <strk> 07/12/21 00:55:24
Modified files:
. : ChangeLog
libbase : LoadThread.h
server/asobj : xml.cpp
Log message:
Don't rely on LoadThread::getBytesTotal, and document why you shouldn't.
CVSWeb URLs:
http://cvs.savannah.gnu.org/viewcvs/gnash/ChangeLog?cvsroot=gnash&r1=1.5235&r2=1.5236
http://cvs.savannah.gnu.org/viewcvs/gnash/libbase/LoadThread.h?cvsroot=gnash&r1=1.17&r2=1.18
http://cvs.savannah.gnu.org/viewcvs/gnash/server/asobj/xml.cpp?cvsroot=gnash&r1=1.59&r2=1.60
Patches:
Index: ChangeLog
===================================================================
RCS file: /sources/gnash/gnash/ChangeLog,v
retrieving revision 1.5235
retrieving revision 1.5236
diff -u -b -r1.5235 -r1.5236
--- ChangeLog 21 Dec 2007 00:39:18 -0000 1.5235
+++ ChangeLog 21 Dec 2007 00:55:23 -0000 1.5236
@@ -1,5 +1,14 @@
2007-12-20 Sandro Santilli <[EMAIL PROTECTED]>
+ * libbase/LoadThread.h: add warnings about getBytesTotal() use
+ and a call for better documenting current behaviour.
+ * server/asobj/xml.cpp (checkLoad): drop the assertion completely
+ and substitute with a check & log_debug. The assertion was
+ about LoadThread::getBytesTotal() returning a number which would
+ be fully available for a singleLoadThread::read() to consume...
+
+2007-12-20 Sandro Santilli <[EMAIL PROTECTED]>
+
* server/asobj/xml.cpp (checkLoad): fixed typo in an assertion.
2007-12-20 Sandro Santilli <[EMAIL PROTECTED]>
Index: libbase/LoadThread.h
===================================================================
RCS file: /sources/gnash/gnash/libbase/LoadThread.h,v
retrieving revision 1.17
retrieving revision 1.18
diff -u -b -r1.17 -r1.18
--- libbase/LoadThread.h 20 Dec 2007 22:31:31 -0000 1.17
+++ libbase/LoadThread.h 21 Dec 2007 00:55:24 -0000 1.18
@@ -90,6 +90,19 @@
long getBytesLoaded();
/// Returns the total size of the file
+ //
+ /// NOTE:
+ /// You can't rely on returned
+ /// value to know the exact size, as network
+ /// server (http/ftp) might return misleading
+ /// information about the size of a resource.
+ ///
+ /// TODO:
+ /// Document current implementation when it
+ /// comes to read past that advertised size
+ /// or short of it... Reading the code is
+ /// discouraging for me (--strk)
+ ///
long getBytesTotal();
/// Check if the load is completed
Index: server/asobj/xml.cpp
===================================================================
RCS file: /sources/gnash/gnash/server/asobj/xml.cpp,v
retrieving revision 1.59
retrieving revision 1.60
diff -u -b -r1.59 -r1.60
--- server/asobj/xml.cpp 21 Dec 2007 00:39:18 -0000 1.59
+++ server/asobj/xml.cpp 21 Dec 2007 00:55:24 -0000 1.60
@@ -17,7 +17,7 @@
// Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
//
-/* $Id: xml.cpp,v 1.59 2007/12/21 00:39:18 strk Exp $ */
+/* $Id: xml.cpp,v 1.60 2007/12/21 00:55:24 strk Exp $ */
#ifdef HAVE_CONFIG_H
#include "config.h"
@@ -471,8 +471,15 @@
size_t xmlsize = lt->getBytesTotal();
boost::scoped_array<char> buf(new char[xmlsize+1]);
size_t actuallyRead = lt->read(buf.get(), xmlsize);
- assert(actuallyRead == xmlsize);
- buf[xmlsize] = '\0';
+ if ( actuallyRead != xmlsize )
+ {
+ // This would be either a bug of LoadThread or
an expected
+ // possibility which lacks documentation (thus
a bug in documentation)
+ //
+ log_debug("LoadThread::getBytesTotal() returned
%d but ::read(%d) returned %d",
+ xmlsize, actuallyRead);
+ }
+ buf[actuallyRead] = '\0';
as_value dataVal(buf.get()); // memory copy here (optimize?)
it = _loadThreads.erase(it);
_______________________________________________
Gnash-commit mailing list
Gnash-commit@gnu.org
http://lists.gnu.org/mailman/listinfo/gnash-commit
