FreeTDS Development Group

[Brian Bruns <camber AT ais.org>]

The patch is (I believe) functional but it's purpose isn't that obvious if 
you don't already know what's going on.  I'd prefer to see an explicit 
check after the 8 byte header has been read (with copious commenting).  I'm 
heading out to a funeral right now, so later tonight (EST) is the earliest 
I can do this unless you want to take a crack at it before then.

We can revisit some of this in 0.61, but for now let's get the minimal 
working change commited.

Brian

On 9 Sep 2002, Frediano Ziglio wrote:

> Il lun, 2002-09-09 alle 14:04, Brian Bruns ha scritto:
> > len is overriden to be header len -8, so at this point i don't know what 
> > it is, could be -8 in which case it'll work, but not too obvious on 
> > reading the code.
> > 
> > Brian
> > 
> 
> Now I understand the real problem...
> 
> x is initially 0.
> This code is executed only if need <= 0, so we do not enter in read
> loop. Then, error is reported if <0 else report 0 (we read an empty
> packet).
> 
> Perhaps we should test if need < 0, close socket and return error (must
> be some garbage...). If packet is empty (need == 0) we can just remove
> this code ( if (x<1){ ...} ).
> 
> Inside loop we test if len == 0, but if len == 0 need == 0 and we do not
> enter in loop...
> 
> On beginning of tds_read_packet we try to read header, if we read only a
> part of packet we just don't close socket but return error. But we lose
> partial information and lost sync, so on next call (if any) it try to
> read header joining partial header with data. Perhaps is best to close
> socket too...
> 
> Looping on goodread for read all packet seem not good... This can only
> happen if read timeout but doing a loop until we read all is the same
> thing of having an infinite timeout for reading packet.
> 
> freddy77
> 
> 
> 
> ---
> You are currently subscribed to freetds as: [camber AT ais.org]
> To unsubscribe, forward this message to $subst('Email.Unsub')
> 
>