summaryrefslogtreecommitdiff
path: root/doc
diff options
context:
space:
mode:
authorRichard Hughes <richard@hughsie.com>2011-04-07 22:43:23 +0100
committerRichard Hughes <richard@hughsie.com>2011-04-07 22:43:37 +0100
commit9ae0fba1816cba725306174f5fc1a35a05bd36ef (patch)
tree3328308881fc23d329b587b0d6490959e288c3ea /doc
parent69ea455a4187970ff178dd08818a6147acf8d452 (diff)
downloadcolord-9ae0fba1816cba725306174f5fc1a35a05bd36ef.tar.gz
Use the embedded profile-id if it is available, else fallback to the file MD5
This should speed up startup as we can use the pre-cooked value rather than using g_compute_checksum_for_data on every profile. This speeds up startup by more than 70% if all the system profiles are fixed.
Diffstat (limited to 'doc')
-rw-r--r--doc/device-and-profile-naming-spec.txt14
1 files changed, 11 insertions, 3 deletions
diff --git a/doc/device-and-profile-naming-spec.txt b/doc/device-and-profile-naming-spec.txt
index d43d67b..8ec13e0 100644
--- a/doc/device-and-profile-naming-spec.txt
+++ b/doc/device-and-profile-naming-spec.txt
@@ -129,10 +129,18 @@ and spaces and does not have to be escaped.
A profile ID is the same for binary identical profiles, even if
duplicated on the filesystem.
-In line with other ICC reccomendations, the complete MD5 file checksum
-is used for the profile ID, e.g.
+In line with other ICC reccomendations, the embedded profile ID (bytes
+84 to 99) is used, after converting it to a hexadecimal encoding.
-Format: icc-{%checksum}
+Format: icc-{%embedded-profile-id}
+Example: icc-3c219b18958837493b586581f8c2a2a4
+
+The ICC do not mandate the embedded profile-id and only recommend it,
+and as such few profiles in the wild use a real value.
+If the embedded profile ID is not set, the complete MD5 file checksum
+should be used instead, e.g.
+
+Format: icc-{%file-checksum}
Example: icc-bd847723f676e2b846daaf6759330624
If a physical profile does not exist (for instance, for embedded printer