A few small bugfixes #490
A few small bugfixes #490
Conversation
Spotted by @pierreganty (#478), thank you
Spotted by @sw17ch, thank you.
Spotted by @sw17ch, thank you.
Previously this gave: ``` ; ./build/bin/re -r literal -G 3 abc ; ./build/bin/re -r literal -G 4 abc abc ; ``` and now -G 3 does construct "abc" (not including the newline, which is a property of the printing, not of the constructed string): ``` ; ./build/bin/re -r literal -G 3 abc abc ; ./build/bin/re -r literal -G 4 abc abc ; ``` and: ``` ; ./build/bin/re -r native -G 3 '^x+$' x xx xxx ``` Spotted by @pierreganty (#478), thank you
| if (ctx->buf_used + ctx->sed[sf->s_id] >= ctx->max_length) { | ||
| LOG(2, "PRUNING due to max length: used:%zu + sed[%d]:%u >= max_length:%zu\n", | ||
| if (ctx->buf_used + ctx->sed[sf->s_id] > ctx->max_length) { | ||
| LOG(2, "PRUNING due to max length: used:%zu + sed[%d]:%u > max_length:%zu\n", |
There was a problem hiding this comment.
@silentbicycle requesting your review for this please, I was unsure about this.
There was a problem hiding this comment.
I think > is correct here -- this means "if the buffer used so far plus what we'd need for the shortest distance to an end state exceeds the available buffer". re -G 5 '^abcde' should produce "abcde" as a valid 5-character match (which is rejected with >=), not counting the trailing \0, but reject re -G 5 '^abcdef'.
This doesn't help for #317, but whatever the solution is there, asserting about it is the wrong thing to do. Spotted by @classabbyamp, thank you
I have extremely broken the iprange example. I am very unsure what's going on with this program.
| if (ctx->buf_used + ctx->sed[sf->s_id] >= ctx->max_length) { | ||
| LOG(2, "PRUNING due to max length: used:%zu + sed[%d]:%u >= max_length:%zu\n", | ||
| if (ctx->buf_used + ctx->sed[sf->s_id] > ctx->max_length) { | ||
| LOG(2, "PRUNING due to max length: used:%zu + sed[%d]:%u > max_length:%zu\n", |
There was a problem hiding this comment.
I think > is correct here -- this means "if the buffer used so far plus what we'd need for the shortest distance to an end state exceeds the available buffer". re -G 5 '^abcde' should produce "abcde" as a valid 5-character match (which is rejected with >=), not counting the trailing \0, but reject re -G 5 '^abcdef'.
-G lengthinconsistent, moreover returns error code. #478, off-by-one forre -G <max-length>and also in the same ticket, wrong exit status forre -G-C opt-level=3does help for performance of rust generated code, and I'd also hoped it'd improve time in CI. But sadly no, that's still dominated by the rustc compile time.