Closed millert closed 9 months ago
This has been confirmed by @enh-google to fix #199
This has been confirmed by @enh-google to fix #199
I can reproduce the problem trivially. I know for a fact that this PR cannot fix #199. as per kernighan's note, it has to do with the size of the gototab.
This has been confirmed by @enh-google to fix #199
I can reproduce the problem trivially. I know for a fact that this PR cannot fix #199. as per kernighan's note, it has to do with the size of the gototab.
well, having tested this patch with the specific input that caused me to raise #199, it definitely does fix that issue. it may introduce other problems, but nothing that our testing finds. do you have an example input that fails with this patch?
We cannot use set_gototab() to reset all the entries for a state, it will leave existing entries as-is. Add a new reset_gototab() function that zeroes the table entries for the specified state. There is no need to reset the goto table immediately after resize_state(), it is already initialized via calloc().