On Sun, Aug 09, 2020 at 08:03:09PM +0800, oldk1331 wrote:
> On Sat, Aug 8, 2020, 9:02 PM Waldek Hebisch <hebi...@math.uni.wroc.pl>
> wrote:
>
> > On Sat, Aug 08, 2020 at 04:59:27PM +0800, oldk1331 wrote:
> > > Why do you think Windows needs different
> > > declarations? Aren't we using cygwin/mingw?
> >
> > For that issue Cygwin, WSL and mingw are really
> > different platforms. WSL emulates Linux, so
> > it should be enough to treat it as Linux. Mingw
> > uses native Windows calls and AFAIK that is
> > different from Linux. In particular AFAIK
> > instead of Unix domain sockets Windows has
> > "local" sockets which offer similar functionality,
> > but require somewhat different code. But
> > for deeper information one should look into
> > Microsoft documentation.
> >
>
> I mean, if I can build fricas in mingw with this
> patch (sockaddr_un) successfully, then can this
> patch be merged? Aka which platform should
> I test for breakage?
What you mean by "this patch". The patch I posted
(adding unused field) should not break anything and
should fix problem that you reported.
In ideal world we should test on all supported systems.
Practically, sockaddr_un is pretty standard Unix things,
so once we get if working on Linux (that is there is
no crude errors) what remains is Windows. As I wrote,
WSL is not a problem, and if Mingw works I would
assume that there are no extra problem with Cygwin.
However, IIUC 32-bit Mingw has differences compared
to 64-bit one, so we should test on both.
Now, comming to "progress", AFAICS the patch I posted
(adding unused field) should allow us to move on with
minimal effort. ATM potential problem with this
patch look rather theortical. However, if problem
would actually show up, then we would be forced to
do deeper (say sockaddr_un) change.
To put it differently, I am trying to avoid breaking
changes, but if I have to choose between having
unfixed bug on Linux and untested code on Windows,
I would choose fixing known bug now (and handling
possible unknown breakage later).
OTOH if I have somewhat unclean code that I expect to
work on all platforms, I prefer this code compared to
cleaner looking but potentialy unportable code.
> > Could you say what exacly you tried? And have you
> > any idea what was so slow? In particular,
> > at what stage build was when you interrupted it?
> >
>
> I'm using msys2. The shell is extremely slow.
> Probably due to anti-virus software (I can't
> disable it, long story). I gave up at ./configure,
> it's 100x slower than Linux. I might try again in
> VM some other time.
I see. Long time to run 'configure' is known problem on
Windows (but 2 hours seems excessive).
--
Waldek Hebisch
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to fricas-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/fricas-devel/20200809131503.GA16378%40math.uni.wroc.pl.
