>>>>>>>>Is there any setting to have this deferred writing turned off, or should I just close and reopen the target file every now and then?
>>>>>>>
>>>>>>>Uses you a buffered file ?
>>>>>>
>>>>>>Mmm... good idea. The source file is buffered. Didn't find any parameters in the fCreate() command that would influence any buffering, but then I could close it immediately after and do a fopen(cTargetFile, 11) (unbuffered, write-only). I'll see what this gives me.
>>>>>
>>>>>Dragan, Don't forget to post the result to me.
>>>>
>>>>fFlush() wasn't even necessary... your idea to use unbuffered LL file operation was exactly what I needed. Works perfect.
>>>
>>>Well Dragan, was what I wanted to feel.
>>>
>>>Certainly that an enigma remains to understand because
>>>FCREATE() doesn't allow to disable the OS buffering.
>>
>>Here's what I did - fcreate()d it, then closed and opened write-only without buffering:
>>

Procedure DoCopy(tcF1, tcF2)
>>*-- dn 26/08/05
>>	Local c, o, N, i, l, cCRS, h1, h2, nSoFar, nChunk
>>
>>*[2005/08/29 09:01:05] dragan - try without buffering, as Fabio suggested
>>	h1=Fopen(tcF1,10)
>>	h2=Fcreate(tcF2)
>>	=fclose(h2)
>*This is the point where the code loses its safety,
>*for some microsecond you lose the control of the file
>* don't close and to use FFLUSH is disgusting but it resolves the problem
>
>>	h2=fopen(tcF2,11)
>>
>>	N=Fseek(h1,0,2)
>>	Fseek(h1,0,0)
>>	nSoFar=N/40
>>	nChunk=nSoFar
>>	Do While !Feof(h1)
>>*		c=Fread(h1, 0x7fff)
>>		Fwrite(h2,Fread(h1, 0x7fff)) && why not this ?
>*
>>		i=Fseek(h1,0,1)
>>		If i>nSoFar
>>			oLog.Log(Transform(i/N*100, "999.99%")+" copied")
>>			nSoFar = nSoFar + nChunk
>>		Endif
>>	Enddo
>>	Fclose(h1)
>>	Fclose(h2)
>

Beg your pardon? ... seems to be you were too fast with the response