kopia lustrzana https://gitlab.com/sane-project/website
59 wiersze
2.6 KiB
HTML
59 wiersze
2.6 KiB
HTML
<!-- received="Fri Jun 5 08:53:13 1998 PDT" -->
|
||
<!-- sent="Fri, 5 Jun 1998 08:49:29 -0700" -->
|
||
<!-- name="Brian Macy" -->
|
||
<!-- email="bmacy@sunshinecomputing.com" -->
|
||
<!-- subject="Eeck.. network protocol" -->
|
||
<!-- id="" -->
|
||
<!-- inreplyto="" -->
|
||
<title>sane-devel: Eeck.. network protocol</title>
|
||
<h1>Eeck.. network protocol</h1>
|
||
<b>Brian Macy</b> (<a href="mailto:bmacy@sunshinecomputing.com"><i>bmacy@sunshinecomputing.com</i></a>)<br>
|
||
<i>Fri, 5 Jun 1998 08:49:29 -0700</i>
|
||
<p>
|
||
<ul>
|
||
<li> <b>Messages sorted by:</b> <a href="date.html#37">[ date ]</a><a href="index.html#37">[ thread ]</a><a href="subject.html#37">[ subject ]</a><a href="author.html#37">[ author ]</a>
|
||
<!-- next="start" -->
|
||
<li> <b>Next message:</b> <a href="0038.html">Kevin Turner: "HP psc/380"</a>
|
||
<li> <b>Previous message:</b> <a href="0036.html">Peter Fales: "SANE options"</a>
|
||
<!-- nextthread="start" -->
|
||
<li> <b>Next in thread:</b> <a href="0082.html">Brian Macy: "Re: Eeck.. network protocol"</a>
|
||
<li> <b>Reply:</b> <a href="0082.html">Brian Macy: "Re: Eeck.. network protocol"</a>
|
||
<li> <b>Reply:</b> <a href="0084.html">Brian Macy: "Re: Eeck.. network protocol"</a>
|
||
<!-- reply="end" -->
|
||
</ul>
|
||
<!-- body="start" -->
|
||
Is it just me or is the SANE network protocol full of race conditions on<br>
|
||
replies? There is now way to identify a reply. It expects the bytes being<br>
|
||
read to be associated with the last sent packet.<br>
|
||
<p>
|
||
screwed. It is possible the remote saned will still send the bytes, even if<br>
|
||
you call SANE_NET_CANCEL. There is no way (that I can tell) that the<br>
|
||
communication link can recover.<br>
|
||
<p>
|
||
Even the sends could incorrectly fail under heavy system loads, in fact I<br>
|
||
believe the watch could cause a process deadlock. If the alarm triggers<br>
|
||
before the actual read/write will the read timeout?<br>
|
||
<p>
|
||
Has anyone thought about a simple frame around these things?<br>
|
||
<p>
|
||
Brian Macy<br>
|
||
<p>
|
||
<p>
|
||
<pre>
|
||
--
|
||
Source code, list archive, and docs: <a href="http://www.mostang.com/sane/">http://www.mostang.com/sane/</a>
|
||
To unsubscribe: echo unsubscribe sane-devel | mail <a href="mailto:majordomo@mostang.com">majordomo@mostang.com</a>
|
||
</pre>
|
||
<!-- body="end" -->
|
||
<p>
|
||
<ul>
|
||
<!-- next="start" -->
|
||
<li> <b>Next message:</b> <a href="0038.html">Kevin Turner: "HP psc/380"</a>
|
||
<li> <b>Previous message:</b> <a href="0036.html">Peter Fales: "SANE options"</a>
|
||
<!-- nextthread="start" -->
|
||
<li> <b>Next in thread:</b> <a href="0082.html">Brian Macy: "Re: Eeck.. network protocol"</a>
|
||
<li> <b>Reply:</b> <a href="0082.html">Brian Macy: "Re: Eeck.. network protocol"</a>
|
||
<li> <b>Reply:</b> <a href="0084.html">Brian Macy: "Re: Eeck.. network protocol"</a>
|
||
<!-- reply="end" -->
|
||
</ul>
|