/appstream-generator

Permalink
Browse files

Don't fail if a suite has no base suite

  • Loading branch information...
1 parent bcece06 commit 224f6e1afaa463b61033a3e31bb32dae0231f566 @ximion committed Feb 11, 2017
Unified Split
Showing with 11 additions and 11 deletions.
  1. +11 −11 src/asgen/engine.d
View
22 src/asgen/engine.d
@@ -183,17 +183,17 @@ public:
// get contents information for packages and add them to the database
auto interestingFound = false;
- // First get the contents (only) of all packages in the base suite.
- logInfo ("Scanning new packages for base suite %s/%s [%s]", suite.baseSuite, section, arch);
- auto baseSuitePkgs = pkgIndex.packagesFor (suite.baseSuite, section, arch);
- foreach (ref pkg; parallel (baseSuitePkgs, 8)) {
- immutable pkid = pkg.id;
-
- string[] contents;
- if (!cstore.packageExists (pkid)) {
- contents = pkg.contents;
- cstore.addContents (pkid, contents);
- logInfo ("Scanned %s for base suite.", pkid);
+ // First get the contents (only) of all packages in the base suite
+ if (!suite.baseSuite.empty) {
+ logInfo ("Scanning new packages for base suite %s/%s [%s]", suite.baseSuite, section, arch);
+ auto baseSuitePkgs = pkgIndex.packagesFor (suite.baseSuite, section, arch);
+ foreach (ref pkg; parallel (baseSuitePkgs, 8)) {
+ immutable pkid = pkg.id;
+
+ if (!cstore.packageExists (pkid)) {
+ cstore.addContents (pkid, pkg.contents);
+ logInfo ("Scanned %s for base suite.", pkid);
+ }
}
}

4 comments on commit 224f6e1

@iainlane
Collaborator

iainlane replied Feb 13, 2017

Ah wtf, look at this commit I have locally on the branch:

commit 1921afa139aa8459d70d51966e010f1093329701
Author: Iain Lane <iain.lane@canonical.com>
Date:   Wed Feb 1 17:54:07 2017 +0000

    engine: Process base suite packages for contents only
    
    We don't need to do any further processing of base suite packages -
    they're only needed to find icons, and we get that from their contents
    alone.
    
    This should stop base suite packages being scanned every time.
    
    If a base suite is later a real suite then everything should continue to
    work.
    
    Fixes #33.

diff --git a/src/asgen/engine.d b/src/asgen/engine.d
index c5ccf93..16abca7 100644
--- a/src/asgen/engine.d
+++ b/src/asgen/engine.d
@@ -182,9 +182,26 @@ public:
 
         // get contents information for packages and add them to the database
         auto interestingFound = false;
+
+        if (!suite.baseSuite.empty) {
+            // First get the contents (only) of all packages in the base suite.
+            logInfo ("Scanning new packages for base suite %s/%s [%s]", suite.baseSuite, section, arch);
+            auto baseSuitePkgs = pkgIndex.packagesFor (suite.baseSuite, section, arch);
+            foreach (ref pkg; parallel (baseSuitePkgs, 8)) {
+                immutable pkid = pkg.id;
+
+                string[] contents;
+                if (!cstore.packageExists (pkid)) {
+                    contents = pkg.contents;
+                    cstore.addContents (pkid, contents);
+                    logInfo ("Scanned %s for base suite.", pkid);
+                }
+            }
+        }
+
+        // And then scan the suite itself - here packages can be 'interesting'
+        // in that they might end up in the output.
         auto pkgs = pkgIndex.packagesFor (suite.name, section, arch);
-        if (!suite.baseSuite.empty)
-            pkgs ~= pkgIndex.packagesFor (suite.baseSuite, section, arch);
         foreach (ref pkg; parallel (pkgs, 8)) {
             immutable pkid = pkg.id;

I guess that I forgot to push this before merging the branch :(

Thanks for fixing! Actually it doesn't affect Ubuntu as I generated the config so that all suites have base suites.
@ximion
Owner

ximion replied Feb 14, 2017

Ah, that explains it! I was doubting myself because I would have sworn that the line was there when I reviewed the patch :D

Thanks for fixing! Actually it doesn't affect Ubuntu as I generated the config so that all suites have base suites.

How does that work? Which base suite does the zesty suite have?
@iainlane
Collaborator

iainlane replied Feb 14, 2017

@ximion
Owner

ximion replied Feb 14, 2017

Jup, that is indeed fine - weird, but fine ^^

I wonder if processing in one go is faster... But given the general slowness of the extraction process, I assume the cost of creating and trashing a few threads is negligible (the rest of the data should have been cached and be reused)
Please sign in to comment.