I'll buy that. build 4530 of legals looks healthy, so let's take it.
jim
On Mar 28, 2007, at 5:06 AM, P T Withington wrote:
Since it's the guts of the compiler, doing a build and running the
demos is probably the best test. I don't think there is much risk
that a correct program will be affected. Mostly I was concerned
about the number of files I had to touch to fix this. The problem
is that in the SWF path, we compile as we go along, so errors are
signaled though one path, whereas in the DHTML path, we accumulate
all the code and then compile, so errors are signaled through a
different path.
The real fix would be to bring the SWF path into conformance with
DHTML, by re-modularizing the compiler, but that is a task for
another day. (I tried to trick Phil R into taking that on a while
back, but he demurred.)
On 2007-03-27, at 23:16 EDT, Jim Grandy wrote:
r4515, yes. r3758, sounds a bit risky. Anything we can do to test
around that change to see if anything was broken?
jim
On Mar 27, 2007, at 8:08 PM, Max Carlson wrote:
I think we should take these...
P T Withington wrote:
Revision: 4515
Bugs Fixed:
LPP-3779 'Misleading Compilation exception syntax error'
Risk: low -- numeric attribute values follow same compilation
path as expressions now
Reward: User-reported bug
Revision: 4518
Bugs Fixed:
LPP-3758 'Java stack trace while compiling malformed JavaScript
code'
Risk: low-medium -- Somewhat complicated changes were required
to the parser/compiler interaction to get proper source
information for errors in DHTML
Reward: User-reported bug
--
Regards,
Max Carlson
OpenLaszlo.org
