Re: [Bug-gnubg] Severe side effects of makebearoff

[Holger]

At 16:28 28.11.2002 +0100, Jim Segrave wrote:
>On Thu 28 Nov 2002 (16:18 +0100), Holger wrote:
>> At 09:09 28.11.2002 +0100, Jim Segrave wrote:
>> >On Wed 27 Nov 2002 (22:56 +0100), Holger wrote:
>> >> Hi,
>> >> 
>> >> There is a severe problem with makebearoff under Windows that probably
>> >> causes wrong creations of the bearoff database.
>> >> 
>> >> When started it returns after about a second (depending on the computer
>> >> speed) and unblocks the shell. In case it's used without a pipe
temp.db is
>> >> then only a few 10k small.
>> >> But the program is still running happily in the background!
>> >> I stumbled over this when I tried to delete the temp file. I could not,
>> >> because of access permissions. Under Win95! So much for M$ error
messages.
>> >> Well, the file was still open and in use.
>> >> I confirmed that makebearoff was still running with WinTop.
>> >> 
>> >> This makes an automatic generation of br1.c impossible, because
>> >> makebearoff1 just takes part of the database as input since
makebearoff is
>> >> still not done. (Maybe you're lucky if you're on a fast machine.)
>> >
>> >Your make is starting parallel processes. That's not wrong, but in
>> >this case it can have the effect you've seen. Something like this
>> >should fix it:
>> 
>> No, this is not caused by make. I've run makebearoff/1 manually from a DOS
>> box. It returns, but keeps running. Really.
>
>I'm very glad I'm running Unix then.

:-) 

However, I'm sure that it's M$ fault. Well, maybe. ;-) It could just be the
implementation of MinGW.

>Nonetheless, using '&&' should force makebearoff to finish before
>makebearoff1 can start.

Hmm, no. && also only operates at shell level and only processes the return
value. Since makebearoff returns without an error, execution is passed on
to the next command. Am I wrong?

Btw, there's _no_ output from either makebearoff nor makebearoff1. At other
times I've seen some.

Regards,

        Holger