The tpdf_time_copy_001 test doesn't initialise time2, and that causes a segfault when we copy into its members.
The API documentation (http://gnupdf.org/manuals/gnupdf-manual/html_node/Managing-Time-Values.html#Managing-Time-Values) says that the destination "Should be a properly created time variable", so I just initialised it. There might be a case for adding some protection (e.g. an assert or error response) to the pdf_time_copy() function. Please apply. Brad --- torture/unit/base/time/pdf-time-copy.c 2009-04-16 22:13:38 +0000 +++ torture/unit/base/time/pdf-time-copy.c 2009-06-13 08:52:12 +0000 @@ -48,6 +48,10 @@ fail_if(status != PDF_OK); fail_if(time1 == NULL); + status = pdf_time_new(&time2); + fail_if(status != PDF_OK); + fail_if(time2 == NULL); + seconds = pdf_i64_new(0x12345678,0x9ABCDEF0); status = pdf_time_set_from_i64(time1, seconds);
