[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: "make test" fails with Cygwin...

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: "make test" fails with Cygwin...
  • Date: Tue, 27 May 2003 13:53:44 -0600
>To: address@hidden
>From: Randy Zagar <address@hidden>
>Subject: Re: 20030527: "make test" fails with Cygwin... 
>Organization: Univ. of Texas
>Keywords: Cygwin

Hi Randy,

> I haven't worked with NetCDF in Cygwin for a while, but I was apparently 
> the first person to get a successful Cygwin version working in 1998 or 
> '99 (I was working at the University of Delaware at the time).
> 
> The "known problems" section mentions that "make test" fails when doing 
> this test:
> 
>      Making `test' in directory [build directory]
>      ..
>      cmp test.nc test_nc.sav
>      test.nc test_nc.sav differ: char 1545, line 5
> 
> If my memory serves me, this problem is due to poor handling of spaces 
> and/or end-of-line sequences when comparing test.nc to test_nc.sav.

I don't think we ever determined the cause of the problem when it was
reported recently, becasue we didn't have a Cygwin platform to test it
on locally.

> This can also be a problem if Cygwin is configured to use the PC-style 
> EOL instead of the traditional Unix standard.  I changed the test script 
> so that it compared checksums (md5, sha, etc...) and this eliminated my 
> problems.

But md5 doesn't ignore differences in spaces or end-of-line sequences,
as far as I understand, and what is being compared is two binary
files.  I don't understand how cmp would report a file difference, but
md5 or sha would say the checksum/digest was the same, unless by an
extremely unlikely coincidence.  I guess I need more details on how
comparing checksums gets around the problem.

We have had a few other users report a similar test failure on Cygwin systems:

 http://www.unidata.ucar.edu/glimpse/netcdf/4180

but subsequent tests showed everything seemed to be working OK despite
this test failure:

 http://www.unidata.ucar.edu/glimpse/netcdf/4181

Another user reported that with his version of Cygwin, everything
built and ran OK "out of the box":

  Running Windows 2000 on a 4000 Mhz Dell laptop, I installed the
  latest cygwin yesterday (4/18/01), and just download netcdf
  3.5.0. Just did

  ./configure
  ./make
  ./make test

  and all worked fine out of the tarball. So, I'm not able to
  reproduce Carl's problem. Consider updating your cygwin environment
  if you haven't done so recently.

So we don't know exactly why some Cygwin installations seem to report
this difference, but apparently the libraries build OK ...

> BTW, do you need someone to maintain a cygwin version of netCDF?  I 
> still get the occasional e-mail about it...

It would be great if you could find time to test our latest beta
release on Cygwin and report back on problems or fixes:

  ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf-beta.tar.Z

It probably has the same problem with "make test" Cygwin that the
3.5.0 release has, but I would be happy to incorporate any fix you
come up with.

Thanks!

--Russ