[Omake] [Ping Jason] Fixing some --divert-* problems.

[Aleksey Nogin]

Jason,

Could you please dbl-check my two recent 1-liner commits (see the two 
attached messages)? Thanks!

A few remaining issues that I see with diversions are all illustrated by 
the rule:

baz:
     echo "About to fail!"
     eprintln(Really fail)
     /bin/fals

When running with --divert-errors, it will first produce (non-diverted)

- build . baz
+ <compute 1 value dependencies>
- build . baz
+ Shell.echo("About to fail!")
About to fail!
- build . baz
+ eprintln(("Really fail"))
Really fail
- build . baz
+ /bin/false

and then repeat (diverted):

- build . baz
+ Shell.echo("About to fail!")
About to fail!
About to fail!
- build . baz
+ eprintln(("Really fail"))
Really fail

The problems are

1) If a rule includes a number of calls to echo/println/eprintln, the 
output of all but the last one of them will be included twice in the 
diversion (but only once in the non-diverted output).

2) The "+/- lines" for the failing step will usually (but somehow not 
always) be missing from the diverted output. Usually this would be 
regardless of whether the final step have produced any output or not.

3) The value dependency stuff does not go into the diverted version.

Aleksey
-------------- next part --------------
An embedded message was scrubbed...
From: nogin at metaprl.org (Aleksey Nogin)
Subject: [Omake] SVN Commit: OMake Build System [0.9.8.x] (Rev. 9792)
Date: Thu, 30 Nov 2006 14:32:45 -0800 (PST)
Size: 2345
Url: http://lists.metaprl.org/pipermail/omake-devel/attachments/20061130/290f5e10/OMakeBuildSystem0.9.8.xRev.mht
-------------- next part --------------
An embedded message was scrubbed...
From: nogin at metaprl.org (Aleksey Nogin)
Subject: [Omake] SVN Commit: OMake Build System [0.9.8.x] (Rev. 9793)
Date: Thu, 30 Nov 2006 14:51:04 -0800 (PST)
Size: 2364
Url: http://lists.metaprl.org/pipermail/omake-devel/attachments/20061130/290f5e10/OMakeBuildSystem0.9.8.xRev-0001.mht