mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benjamin Bannier <benjamin.bann...@mesosphere.io>
Subject Re: Review Request 65322: Made `numify` less dependent on `boost::lexical_cast` failure behavior.
Date Thu, 25 Jan 2018 14:32:13 GMT


> On Jan. 25, 2018, 2:18 p.m., Benno Evers wrote:
> > 3rdparty/stout/include/stout/numify.hpp
> > Lines 34 (patched)
> > <https://reviews.apache.org/r/65322/diff/1/?file=1946053#file1946053line34>
> >
> >     Its not your comment, but this point sounds like a red herring to me: We're
always parsing strings here, so the it shouldn't matter what the rules for permissible floating-point
literals are.

I agree, it also does not matter what some standard has to say about hex number formats; we
are parsing a string here. I trimmed this comment.


> On Jan. 25, 2018, 2:18 p.m., Benno Evers wrote:
> > 3rdparty/stout/include/stout/numify.hpp
> > Lines 43 (patched)
> > <https://reviews.apache.org/r/65322/diff/1/?file=1946053#file1946053line43>
> >
> >     Why 'maybe'? I'd suggest
> >     
> >     ```
> >     bool isHex = strings::startsWith(s, "0x") || ...;
> >     ```

I used `maybe` since this is only a heuristic which would e.g., match a string like `0x XYZ`
as well which is not a literal of a hex number.


- Benjamin


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/65322/#review196239
-----------------------------------------------------------


On Jan. 25, 2018, 3:32 p.m., Benjamin Bannier wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/65322/
> -----------------------------------------------------------
> 
> (Updated Jan. 25, 2018, 3:32 p.m.)
> 
> 
> Review request for mesos, Benno Evers, Till Toenshoff, and Cong Wang.
> 
> 
> Bugs: MESOS-8484
>     https://issues.apache.org/jira/browse/MESOS-8484
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Previously we relied on Boost to never successfully parse hexadecimal
> numbers, and only performed rejection of hexadecimal floating point
> numbers before entering our custom parser for hexadecimal numbers.
> This was not reliable as on some platforms `boost::lexical_cast` ended
> up successfully parsing literals of hexadecimal floating point numbers
> leading us to never rejecting hexadecimal floating point numbers.
> 
> This patch moves the rejection of hexadecimal floating point numbers
> to a point before attempting a parse with `boost::lexical_cast`. While
> this decouples us in some regards from Boost's parsing behavior, we
> also might end up performing more work than needed. Should this become
> a concern we should revisit and optimize the implementation here.
> 
> 
> Diffs
> -----
> 
>   3rdparty/stout/include/stout/numify.hpp 6db9a78145b7e90cc975786ca83f7acb7fdc3e0a 
> 
> 
> Diff: https://reviews.apache.org/r/65322/diff/2/
> 
> 
> Testing
> -------
> 
> `make check` on a collection of Linux, macos and Windows setups in internal CI.
> 
> 
> Thanks,
> 
> Benjamin Bannier
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message