fix: check vaild frist when relative_pos_range close #94

[yihong0618]

as title, close #94

[Unparalleled-Calvin]

Hi, yihong. Thank you for your contribution! The keyword unsafe is used here to notice users that there might be potential issues when using this API including:

• The two spans may come from different source files which makes the result meaningless.
• The overflow issues.

We can write code to check both issues and ensure that the program won't crash. However, the bugs from the caller function won't be detected. Considering that the bugs are conner cases, I think adding a visible notice will make sense?

[yihong0618]

Hi, yihong. Thank you for your contribution! The keyword unsafe is used here to notice users that there might be potential issues when using this API including:

• The two spans may come from different source files which makes the result meaningless.
• The overflow issues.

We can write code to check both issues and ensure that the program won't crash. However, the bugs from the caller function won't be detected. Considering that the bugs are conner cases, I think adding a visible notice will make sense?

Hi thank you for the comments

IMO for this kind of error the better way is do not panic and stop the world?
a warning message and drop the unsafe maybe better?

---

can you try using RAPX for greptime, the error happens every time when using -ops for greptime, maybe there is a better solution for this issue

[Unparalleled-Calvin]

Hi yihong. I have checked the code and figured out the reason that caused this bug. In short, the sub_span passed into the function locates at a macro code block which is outside the span.

An example:

macro_rules! foo{
    ... // the sub_span locates at here
}

fn bar() { // span
    foo!(...) // unfolded while compiling
}

Then the previous code computes relative_pos_range(span, sub_span) which causes an overflow bug.

Thank you for your contribution!

[yihong0618]

Hi yihong. I have checked the code and figured out the reason that caused this bug. In short, the sub_span passed into the function locates at a macro code block which is outside the span.

An example:

macro_rules! foo{
    ... // the sub_span locates at here
}

fn bar() { // span
    foo!(...) // unfolded while compiling
}

Then the previous code computes relative_pos_range(span, sub_span) which causes an overflow bug.

Thank you for your contribution!

Thank you for the info, learned that