[LOW] STR #4075: PPD: Message Processing Bottlenecks
Chad Hulbert <Chad.Hulbert-Y/[email protected]>
| Newsgroups | gmane.comp.printing.cups.bugs |
|---|---|
| Message-ID | <[email protected]> |
DO NOT REPLY TO THIS MESSAGE. INSTEAD, POST ANY RESPONSES TO THE LINK BELOW. [STR New] The currently recommended approach for encoding binary data in PPD files through PPD: messages from a filter suffers serious performance issues when the amount of data becomes large. Even more so when the PPD file is large. Each PPD: message causes the PPD file to be read by line and written to a copy. If a filter emits 10s or 100s of messages this can result in 10s or 100s of megabytes of disk access, which can make the scheduler non-responsive. On OS X, this can lock the application and deny service to concurrent users. While concise encoding, compression and multiple attributes per PPD: message help reduce the impact, the problem remains unsolved. Link: http://www.cups.org/str.php?L4075 Version: 1.6-current