Re: [tree-ssa] Insert on edge fix

[law at redhat dot com]

In message <Pine.LNX.4.44.0306201248340.25569-100000@ops2.one-point.com>, Jeff 
Sturm writes:
 >On 20 Jun 2003, Andrew MacLeod wrote:
 >> 	* tree-cfg.c (handle_switch_split):  Revert last change.
 >
 >Merging this with Jeff's handle_switch_split changes from yesterday and
 >my gcj frontend work, I get:
 >
 >        === libstdc++-v3 Summary ===
 >
 ># of expected passes        1853
 ># of unexpected failures    1
 ># of unexpected successes   1
 >
 >        === libmudflap Summary ===
 >
 ># of expected passes        4
 ># of unexpected failures    268
 >
 >        === libjava Summary ===
 >
 ># of expected passes        3018
 ># of unexpected failures    1
 ># of expected failures      16
 ># of untested testcases     12
 >
 >        === gcc Summary ===
 >
 ># of expected passes        22430
 ># of unexpected failures    12
 ># of unexpected successes   6
 ># of expected failures      79
 ># of unresolved testcases   1
 ># of unsupported tests      101
 >
 >        === g++ Summary ===
 >
 ># of expected passes        8557
 ># of unexpected failures    13
 ># of unexpected successes   1
 ># of expected failures      68
 ># of unsupported tests      15
 >
 >        === objc Summary ===
 >
 ># of expected passes        1156
 >
 >        === g77 Summary ===
 >
 ># of expected passes        1734
 ># of unexpected failures    1
 ># of untested testcases     1
 >
 >
 >Not bad... though I'm suspicious of the g++ failures.  Significantly,
 >libjava builds with no extra backend hacks.
 >
 >It's up to you guys, but I'd really, really like to get the gcj stuff in
 >soon so I can stop playing catchup.  As the tree stands today (i.e.
 >without Andrew's reverted patch) it won't build.
I strongly recommend not reverting Andrew's patch.  Andrew will be back
online Monday working regularly and I'm sure he'll give the problem
prompt attention.

jeff