On November 26, 2003 at 18:53, knok(_at_)daionet(_dot_)gr(_dot_)jp wrote:
According to gdb, it's in the following line that the reason why
namazu couldn't work appropriately:
val = nmz_do_searching(token, val);
(nmz/parser.c: factor())
val is an structure (NmzResult). In nmz_do_searching() function, the
value of val seems handled appropriately, but the return value was not
effected, therefore val.num is still 0 after the line.
I maed a patch for the issue. But 2 tests were still failed.
Is there a concern that namazu code be reentrant? If so, it
appears that your patch will break this.
I do find it interesting that the namazu code passes structures
around by copy a lot. Is there a reason for this?
--ewh