From e009537167861477896703935148daaf2b2dc31f Mon Sep 17 00:00:00 2001
From: Bruce Allen <bruce.allen@aei.mpg.de>
Date: Thu, 30 Dec 2004 04:16:00 +0000
Subject: [PATCH] When creating large numbers of WU which have large input
 files, the cost of computing the md5 sums becomes expensive.  Inserted a
 simple caching mechanism into the utility function process_wu_template(). 
 For a given filename FILENAME, this looks for a file called FILENAME.md5,
 which contains the md5sum and length.  If the file exists, these values are
 used.  If not, such a file is created.  Since (I think) David is planning to
 have a database for files, this mechanism will eventually be obsolete. Right
 now it disabled by default.  Use -DBOINC_CACHE_MD5 to enable it.

svn path=/trunk/boinc/; revision=4965
---
 checkin_notes       | 16 ++++++++++++++++
 tools/backend_lib.C | 40 ++++++++++++++++++++++++++++++++++++----
 2 files changed, 52 insertions(+), 4 deletions(-)

diff --git a/checkin_notes b/checkin_notes
index 20b6345b02..20b7ed9d5e 100755
--- a/checkin_notes
+++ b/checkin_notes
@@ -21995,3 +21995,19 @@ David  29 Dec 2004
 
     api/
         x_opengl.C
+
+Bruce  30 Dec 2004
+
+    - When creating large numbers of WU which have large input files,
+      the cost of computing the md5 sums becomes expensive.  Inserted
+      a simple caching mechanism into the utility function
+      process_wu_template().  For a given filename FILENAME, this looks
+      for a file called FILENAME.md5, which contains the md5sum and
+      length.  If the file exists, these values are used.  If not, such
+      a file is created.  Since (I think) David is planning to have a
+      database for files, this mechanism will eventually be obsolete.
+      Right now it disabled by default.  Use -DBOINC_CACHE_MD5 to
+      enable it.
+
+    tools/
+        backend_lib.C
diff --git a/tools/backend_lib.C b/tools/backend_lib.C
index f9a7ecfaae..053879c4e5 100644
--- a/tools/backend_lib.C
+++ b/tools/backend_lib.C
@@ -109,11 +109,43 @@ static int process_wu_template(
                         boinc_copy(top_download_path,path);
                     }
 
-                    retval = md5_file(path, md5, nbytes);
-                    if (retval) {
-                        fprintf(stderr, "process_wu_template: md5_file %d\n", retval);
-                        return retval;
+                    retval=0;
+#ifdef BOINC_CACHE_MD5
+                    // see checkin-notes Dec 30 2004
+                    {
+                        // look for file named FILENAME.md5 containing
+                        // md5sum and length
+                        FILE *fp;
+                        char md5name[512];
+                        sprintf(md5name, "%s.md5", path);
+                        if ((fp=fopen(md5name, "r"))) {
+                            // found cached file info!
+                            if (2==fscanf(fp, "%s %lf", md5, &nbytes))
+                              retval=1; // indicates sucess getting cached info
+                            fclose(fp);
+                        }
                     }
+#endif
+                    if (!retval) {
+                        retval = md5_file(path, md5, nbytes);
+                        if (retval) {
+                            fprintf(stderr, "process_wu_template: md5_file %d\n", retval);
+                            return retval;
+                        }
+#ifdef BOINC_CACHE_MD5
+                        else {
+                            // Write file FILENAME.md5 containing md5sum and length
+                            FILE *fp;
+                            char md5name[512];
+                            sprintf(md5name, "%s.md5", path);
+                            if ((fp=fopen(md5name, "w"))) {
+                                fprintf(fp,"%s %.15e\n", md5, nbytes);
+                                fclose(fp);
+                            }
+                        }
+#endif
+                    } // (!retval)
+
                     dir_hier_url(
                         infiles[file_number], config.download_url,
                         config.uldl_dir_fanout, url