Bug 1623010 - Separate out the diffing of generated-files. r=dmajor

The diff jobs currently take a Firefox package and compares against the
one from another build. When that fails, it also makes the diff of the
generated files, which can be useful when there are differences, but
can also be useful on its own. Also, because sometimes, like right now,
there are differences in generated files that have no impact on Firefox
itself, when differences do show up for Firefox, the differences in
generated files are added noise that sheriffs can't work around.

Differential Revision: https://phabricator.services.mozilla.com/D67113

--HG--
extra : moz-landing-system : lando

taskcluster/ci/diffoscope/kind.yml

@@ -66,6 +66,14 @@ jobs:
         fail-on-diff: true
         run-on-projects: ['try', 'mozilla-central']  # Do not run linux32 on autoland, Bug 1599197
 
+    reproducible-linux32-generated-files:
+        symbol: Linux32g
+        artifact: target.generated-files.tar.gz
+        new: build-linux-reproduced/opt
+        original: build-linux-shippable/opt
+        fail-on-diff: true
+        run-on-projects: ['try', 'mozilla-central']  # Do not run linux32 on autoland, Bug 1599197
+
 
 # Make a task for each diff we might want. The following are just examples,
 # Both original and new can point to builds from the full set of tasks or

taskcluster/docker/diffoscope/get_and_diffoscope

@@ -34,6 +34,10 @@ case "$ORIG_URL" in
 		$MOZ_FETCHES_DIR/dmg/hfsplus $i.hfs extractall / $i
 	done
 	;;
+*)
+	ARTIFACT=$(basename "${ORIG_URL}")
+	curl -L "$ORIG_URL" > "a/${ARTIFACT}"
+	curl -L "$NEW_URL" > "b/${ARTIFACT}"
 esac
 
 case "$ORIG_URL" in
@@ -124,26 +128,6 @@ then
 	# Ok
 	:
 else
-	# The builds differ, let's check the generated files.
-	if curl -L -f "${ORIG_URL%/*}/target.generated-files.tar.gz" > a.generated-files.tar.gz && \
-	   curl -L -f "${NEW_URL%/*}/target.generated-files.tar.gz" > b.generated-files.tar.gz
-	then
-		HAVE_GENERATED_FILES=1
-	else
-		HAVE_GENERATED_FILES=
-	fi
-
-	if test -z "$HAVE_GENERATED_FILES" || diffoscope \
-		--html generated-files.diff.html \
-		--text generated-files.diff.txt \
-		--progress \
-		--output-empty \
-		a.generated-files.tar.gz b.generated-files.tar.gz
-	then
-		report_error diff
-	else
-		report_error generated-files.diff
-	fi
-
+	report_error diff
 	$POST
 fi

taskcluster/taskgraph/transforms/diffoscope.py

@@ -50,6 +50,11 @@ diff_description_schema = Schema({
     # Fail the task when differences are detected.
     Optional('fail-on-diff'): bool,
 
+    # What artifact to check the differences of. Defaults to target.tar.bz2
+    # for Linux, target.dmg for Mac, target.zip for Windows, target.apk for
+    # Android.
+    Optional('artifact'): text_type,
+
     # Whether to unpack first. Diffoscope can normally work without unpacking,
     # but when one needs to --exclude some contents, that doesn't work out well
     # if said content is packed (e.g. in omni.ja).
@@ -76,6 +81,7 @@ def fill_template(config, tasks):
         deps = {}
         urls = {}
         previous_artifact = None
+        artifact = task.get('artifact')
         for k in ('original', 'new'):
             value = task[k]
             if isinstance(value, text_type):
@@ -100,7 +106,9 @@ def fill_template(config, tasks):
                 deps[index] = 'index-search-' + index
                 dep_name = index
                 os_hint = index.split('.')[-1]
-            if 'linux' in os_hint:
+            if artifact:
+                pass
+            elif 'linux' in os_hint:
                 artifact = 'target.tar.bz2'
             elif 'macosx' in os_hint:
                 artifact = 'target.dmg'
@@ -139,8 +147,6 @@ def fill_template(config, tasks):
                 } for f in (
                     'diff.html',
                     'diff.txt',
-                    'generated-files.diff.html',
-                    'generated-files.diff.txt',
                 )],
                 'env': {
                     'ORIG_URL': urls['original'],