On Fri, Aug 16, 2013 at 6:35 PM, Richard Hipp <[email protected]> wrote:
> We could add a check at the top of the manifest parser loop to check every > card using strcmp(). But that would slow down processing in a performance > critical section of the code, so I left that out. > i figured that was the reason and didn't feel compelled to go add _quite_ that level of validation to my ports. > If you do find a case there cards are not checked for strict ordering, > then that is a bug that needs fixing. > No, no - this has all been curiosity on my part, not a bug report. i've spent most of today downloading raw manifests and running them through my parser/generator APIs to ensure 100% fidelity and equivalent consistency checking: [stephan@host:~/cvs/fossil/f2]$ ./test-mf resources/mf-cherrypick-beabf372b5.txt Parsing this manifest: resources/mf-cherrypick-beabf372b5.txt MARKER: test-mf.c:42:test_parse_1(): Manifest type=MANIFEST MARKER: test-mf.c:46:test_parse_1(): Trying to re-calculate R-card: original=[e678ad4e8726ea23981e00fb9512cf4b] MARKER: test-mf.c:53:test_parse_1(): Re-calculated R-card: [e678ad4e8726ea23981e00fb9512cf4b] Round-trip re-generated manifest (type=MANIFEST) from input file: Rather large - not dumping to console. Dumping mf to file [mf.out] [stephan@host:~/cvs/fossil/f2]$ diff resources/mf-cherrypick-beabf372b5.txt mf.out [stephan@host:~/cvs/fossil/f2]$ And it's pretty simple to do: http://fossil.wanderinghorse.net/repos/f2/index.cgi/artifact/99017fa9551ad0419aa5838cb0b8041b4c6745a1 :) -- ----- stephan beal http://wanderinghorse.net/home/stephan/ http://gplus.to/sgbeal
_______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
