On Sat, 2014-09-27 at 03:57 +0000, Miklos Vajna wrote:
> New crashtest update available at 
> http://dev-builds.libreoffice.org/crashtest/e29862bcc509916e7fe3e9b2f43b0672c49b7f6e/

Import crash improvement is encouraging anyway and I fixed another one
or two .docx imports today. As we were talking about this in the ESC
recently I'd just like to point out that 
a) the tests are done with a dbgutil version of libreoffice so asserts
are fails.
b) the "doc" column are things that are described as the msword
mime-type in bugzillas and pretty must all the fails are actually
miscategorized rtf not doc
c) in my local get-bugzilla-attachments-by-mimetype cache we have loads
of weird-ass obscure formats with no examples at all, e.g. "602", "fb2",
etc.
d) it should also be possible to determine from the documents that we do
have the approximate smallest subset of documents required to fulfill
the maximum code-coverage that the full set covers. e.g. with something
like intel's pin or some other approach. That would give the optimal
seed set for format fuzzing.

C.

_______________________________________________
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Reply via email to