Merge branch 'master' of github.com:redpony/cdec

4 files changed, 26 insertions, 6 deletions

diff --git a/SConstruct b/SConstruct
index c21d85d5..41c60178 100644
--- a/SConstruct
+++ b/SConstruct
@@ -29,7 +29,8 @@ env = Environment(PREFIX=GetOption('prefix'),
 boost = GetOption('boost')
 if boost:
    print 'Using Boost at {0}'.format(boost)
-   env.Append(CPPPATH=boost+'/include',
+   env.Append(CCFLAGS='-DHAVE_BOOST',
+              CPPPATH=boost+'/include',
 	      LIBPATH=boost+'/lib')
 
 if GetOption('efence'):
@@ -45,6 +46,8 @@ if glc:
    env.Append(CCFLAGS='-DHAVE_GLC',
 	      CPPPATH=[glc, glc+'/cdec'])
    srcs.append(glc+'/string_util.cc')
+   srcs.append(glc+'/sys_util.cc')
+   srcs.append(glc+'/debug.cc')
    srcs.append(glc+'/feature-factory.cc')
    srcs.append(glc+'/cdec/ff_glc.cc')
 
diff --git a/environment/LocalConfig.pm b/environment/LocalConfig.pm
index d136610d..7b3d950c 100644
--- a/environment/LocalConfig.pm
+++ b/environment/LocalConfig.pm
@@ -37,7 +37,7 @@ my $CCONFIG = {
     'HOST_REGEXP' => qr/^(thor|tyr)\.inf\.ed\.ac\.uk$/,
   },
   'Blacklight' => {
-    'HOST_REGEXP' => qr/^(blacklight.psc.edu|bl1.psc.teragrid.org|bl0.psc.teragrid.org)$/,
+    'HOST_REGEXP' => qr/^(tg-login1.blacklight.psc.teragrid.org|blacklight.psc.edu|bl1.psc.teragrid.org|bl0.psc.teragrid.org)$/,
     'QSubMemFlag' => '-l pmem=',
   },
   'Barrow/Chicago' => {
diff --git a/klm/util/string_piece.hh b/klm/util/string_piece.hh
index e48ce3d9..2583db5e 100644
--- a/klm/util/string_piece.hh
+++ b/klm/util/string_piece.hh
@@ -48,7 +48,10 @@
 #ifndef BASE_STRING_PIECE_H__
 #define BASE_STRING_PIECE_H__
 
-#include "util/have.hh"
+//Uncomment this line if you use ICU in your code.  
+//#define HAVE_ICU
+//Uncomment this line if you want boost hashing for your StringPieces.
+//#define HAVE_BOOST
 
 #ifdef HAVE_BOOST
 #include <boost/functional/hash/hash.hpp>
diff --git a/vest/dist-vest.pl b/vest/dist-vest.pl
index 80d2471e..789b5b14 100755
--- a/vest/dist-vest.pl
+++ b/vest/dist-vest.pl
@@ -289,9 +289,23 @@ while (1){
 	my $cmd = "$pcmd $decoder_cmd 2> $decoderLog 1> $runFile";
 	print STDERR "COMMAND:\n$cmd\n";
 	check_bash_call($cmd);
-        my $num_hgs = check_output("ls $dir/hgs/*.gz | wc -l");
-        print STDERR "NUMBER OF HGs: $num_hgs\n";
-	die "Dev set contains $devSize sentences! Decoder failure?\n" if ($devSize != $num_hgs);
+        my $num_hgs;
+        my $num_topbest;
+        my $retries = 0;
+	while($retries < 5) {
+	    $num_hgs = check_output("ls $dir/hgs/*.gz | wc -l");
+	    $num_topbest = check_output("wc -l < $runFile");
+	    print STDERR "NUMBER OF HGs: $num_hgs\n";
+	    print STDERR "NUMBER OF TOP-BEST HYPs: $num_topbest\n";
+	    if($devSize == $num_hgs && $devSize == $num_topbest) {
+		last;
+	    } else {
+		print STDERR "Incorrect number of hypergraphs or topbest. Waiting for distributed filesystem and retrying...\n";
+		sleep(3);
+	    }
+	    $retries++;
+	}
+	die "Dev set contains $devSize sentences, but we don't have topbest and hypergraphs for all these! Decoder failure? Check $decoderLog\n" if ($devSize != $num_hgs || $devSize != $num_topbest);
 	my $dec_score = check_output("cat $runFile | $SCORER $refs_comma_sep -l $metric");
 	chomp $dec_score;
 	print STDERR "DECODER SCORE: $dec_score\n";