cygwin: gap.cputime() does not work

[williamstein]

sage: gap.cputime()
---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)

/home/wstein/sage-4.4.3/<ipython console> in <module>()

/home/wstein/sage-4.4.3/local/lib/python2.6/site-packages/sage/interfaces/gap.pyc in cputime(self, t)
    429         else:
    430             self.eval('_r_ := Runtimes();')
--> 431             r = sum(eval(self.eval('[_r_.user_time, _r_.system_time, _r_.user_time_children, _r_.system_time_children]')))
    432             return r/1000.0
    433 

/home/wstein/sage-4.4.3/local/lib/python2.6/site-packages/sage/interfaces/gap.pyc in <module>()

NameError: name 'fail' is not defined
sage: 

CC: @jpflori @dimpase

Component: porting: Cygwin

Reviewer: Jean-Pierre Flori, Karl-Dieter Crisman

Issue created by migration from https://trac.sagemath.org/ticket/9164

[kcrisman]

comment:1

Hey, is this related to the mysterious comment

sage: v, t = qsieve(n, time=True)   # uses the sieve    (optional: time doesn't work on cygwin) 

in sage/interfaces/qsieve.py? And does time now work on Cygwin?

[jpflori]

comment:2

Dont think so.

gap.cputime() works on both my systems.
It is even so nice it reports twice as much on the 32 bits than on the 64 bits...

So let's close this one.

Nonetheless the other qsieve examples do not work, this should be treated elsewhere.

[jpflori]

comment:3

Replying to @jpflori:

Nonetheless the other qsieve examples do not work, this should be treated elsewhere.

Or not:
http://comments.gmane.org/gmane.os.cygwin/106331
Although the time bash builtin works, there is no time command under Cygwin (nor in any package), so either we should modify the qsieve code (what will have to be done anyway after #12173 gets in and we get rid of qsieve which will then be obsoleted), or live with such code being optional.

[jpflori]

comment:4

My bad, I wrongly used cygcheck.
There is http://cygwin.com/packages/time/
So either we make time a prereq which I would not advocate for, or rather test at runtime if the real time executable is available.
But thats for another ticket anyway (unless I get my hand on a Mac, fix #12173 and remove qsieve by that time).

[kcrisman]

comment:5

JP, see #14184. What do you think?

[jdemeyer]

comment:6

Replying to @jpflori:

Although the time bash builtin works, there is no time command under Cygwin (nor in any package)

This has nothing to do with Cygwin. My Gentoo Linux system doesn't have a time command either, it does have the time keyword (to be pedantic: it's a keyword, not a builtin) in bash.

so [...] we should modify the qsieve code

Exactly, see #14202.

[kcrisman]

Reviewer: Jean-Pierre Flori, Karl-Dieter Crisman

[kcrisman]

comment:7

After a rebase, this works! Awesome.