Bug #14220: WEBrick changes - failures on MSWIN, MinGW

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 2.5.0dev (2017-12-22 trunk 61404) [x64-mingw32]
• Backport: 2.3: UNKNOWN, 2.4: UNKNOWN

[MSP-Greg (Greg L)]

@normalperson (Eric Wong)

Eric,

Hopefully this isn't controversial

What? If someone thinks it is, maybe they can volunteer to maintain everything you said you'd help with... Anyway, thanks for all the time you've put in already.

Windows is kind of crashing & burning with the recent changes to WEBrick, etc. I haven't really looked at them, but thought the logs might be helpful. Attached, both have 'retry' info above the normal failure summary info. I'm -0600, so it's 23:30...

Thanks again, Greg

• Status changed from Open to Closed

Applied in changeset trunk|r61409.

---

webrick/httpservelet/cgi_runner: avoid IO#reopen on pathname

IO#reopen seems to behave strangely on Win32 based on the logs
posted by Greg for [Bug #14420]. Lets try to fix the issue by
using File.open explicitly (but not Kernel#open).

• lib/webrick/httpservelet/cgi_runner: use File.open explicitly
  [Bug #14220]

[normalperson (Eric Wong)]

[e-mail] wrote:

Eric,

Hopefully this isn't controversial

What? If someone thinks it is, maybe they can volunteer to
maintain everything you said you'd help with... Anyway,
thanks for all the time you've put in already.

I guess that is in response to [Misc #14210] and [Misc #14211]?

shrug I assume a lot of people don't like my stance against
proprietary services and my refusal to deal with graphical stuff;
and don't like my work as a result.

Anyways you're welcome :>

Windows is kind of crashing & burning with the recent changes
to WEBrick, etc. I haven't really looked at them, but thought
the logs might be helpful. Attached, both have 'retry' info
above the normal failure summary info. I'm -0600, so it's
23:30...

Oops, I made r61409 to try to fix this. Hopefully it's OK...

Otherwise we could try to revert both r61409 and r61398 and
add a comment explaining the use of Kernel#open...

I also wonder if IO#reopen would work for you without the "w"
arg. I know the rb_freopen function is completely different
on Win32, so it's likely there's something strange there...

Maybe @usa (Usaku NAKAMURA) can chime in...

[normalperson (Eric Wong)]

Eric Wong [e-mail] wrote:

[e-mail] wrote:

Windows is kind of crashing & burning with the recent changes
to WEBrick, etc. I haven't really looked at them, but thought
the logs might be helpful. Attached, both have 'retry' info
above the normal failure summary info. I'm -0600, so it's
23:30...

Oops, I made r61409 to try to fix this. Hopefully it's OK...

Ping. Can you confirm r61409 is alright? Thanks.