SV: SV: egcs-2.92.33 19981226 testsuite on i686-pc-cygwin32

N8TM@aol.com N8TM@aol.com
Sun Jan 31 23:52:00 GMT 1999


In a message dated 1/9/99 1:21:58 AM Pacific Standard Time,
c.christian.joensson@telia.com writes:

<< make[3]: *** Warning:  Clock skew detected.  Your build may be incomplete.
  >>
This warning is normal, both on NT and W95. I had one yesterday on NT. It
seems that the lowest order bit of the time call return value is random.  I've
never seen it to cause problems on cygwin.  It does signal a problem where you
are building across a network, and the file time clocks of the various
machines may be several minutes out.  Maybe we could report this as a gnu-
win32 bug, suggesting someone familiar with this timer modify it to avoid the
single bit back ticks.  It was reported several times under B19, and it seems
not to have changed.

>>I did install the Cugwin B20.1 and on top of that, Mumit's egcs-1.1.1 for
B20.
Then, tried the dev-src, coudn't make it in the first place. 

As I mentioned before, I haven't attempted to rebuild the entire dev-src using
egcs-1.1.1.  I rebuilt it entirely intact except for replacing make with
make-3.77, then installed egcs-1.1.1 and rebuilt the tools which were giving
trouble (patch, expect, tcl).  Certainly, it would be interesting to know
whether the dev-src is entirely compatible with egcs-1.1.1 with Mumit's
patches.   Mumit's patches are definitely required to avoid some linking
failures in g++; I don't know whether the C patches would have any effect on
building dev-src.  I've had bad experiences before, trying to add tools to the
integrated egcs-type build structures, so I would look first for a way to
attempt to get it to use the installed compiler and not try to build its own.

>>I realized that I would
perhaps get away with the thing using egcs' infrastructure dejagnu-19981026

I've gone back and forth over the various dejagnu distributions, and the one
in dev-src works as well as any, provided that expect and tcl are built with
the most reliable compiler available.  I find a slight improvement after
installing Okhapkin's latest coolview .dll from ftp.franken.de.  It is
designed to work with bash installed as /bin/sh but the apparent redundancy of
continuing to use b20.1 ash as /bin/sh together with coolview must be all to
the good, as long as we are running software which doesn't specifically
require bash.

>>Oh, an other story, at work I installed Cygwin B20.1 under MetaFrame from
Citrix. The installations seems to work. Should one consider it installed
under
Win NT4.0 or should one consider it a new OS (it is after all *not* Win
NT4.0(sp3)
but rather Win NT4.0 terminal server edition...)?

I suppose that it's worth mentioning any time you describe behaviors which
might be affected by which version of Windoze you have.  Certainly, SP changes
are suspected of having effects, USB has effects under W95.
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".



More information about the Cygwin mailing list