Question

我已经实现了解析大文件的概念，如下所示，但似乎不正确。我面对代码的问题很少，一些变量无法访问。

    #ALL the variable in Complete CAPS are global variable
    TLog("MSG",1,"Parent process $$");
    TLog("MSG",4,"Creating child process for $$");
    my $MAX_FORK       = 2;
    my $forkCount      = 0;
    my $processCounter = 0;
    my @childId   = ();

    foreach my $fileNameFasta (@{$ref_array_file}) {

        my $pid = fork();


        if ( $pid ) {

            TLog("MSG",1,"child process created : $pid");
            push @childId,$pid;

            $forkCount++;
        }
        elsif ( $pid == 0 ) {

            my $outputFile = $STAT_FILE;
            my $pidLocal   = $childId[$processCounter]; #Use of unintialized variable

            $outputFile =~s/\d{1,}\.txt$/$pidLocal\.txt/og; #hence naming of all ouput file are same                
            TLog("MSG",1,"For $pidLocal Creating output file for stat : $outputFile");

            open my $outputfh,'>',$outputFile;
            GenerateTupleCountFile($outputfh,$fileNameFasta);    
            close  $outputfh;


            TLog("MSG",5,"Calculation completed for $pidLocal");
            TLog("MSG",5,"Plz check the $outputFile");
            $processCounter++;

            exit(0);

        }

        if ( $forkCount >= $MAX_FORK ) {
            foreach (@childId) {
                   my $tmp = waitpid($_, 0);
                   TLog("MSG",5,"Process completed for with pid $tmp");
            }
        }
    }
}

如果我是冤枉，请给我正确的指示。

Answer 1

一旦进行了分叉，父母和孩子就会自主。

在@childID之后，你永远不会在fork数组中设置任何内容，到那时，孩子知道在那里写的内容为时已晚。您需要在孩子中使用getpid()，或者使用神奇变量$$，或者（如果您使用过use English '-no_match_vars';）$PID或$PROCESS_ID。父级永远不会增加$processCounter。

同样，孩子会增加$processCounter的副本，但这不会影响父母的变量。

perl forking实现了大量的文件解析

1 个答案: