From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Garzik Subject: Re: [PATCHv2] cld: use XDR for all messages Date: Tue, 09 Feb 2010 20:33:55 -0500 Message-ID: <4B720D03.3020101@garzik.org> References: <1263577788-2081-1-git-send-email-cmccabe@alumni.cmu.edu> <4B7128FB.90600@garzik.org> <4B712D0D.5060408@garzik.org> <20100209161306.5e312b28@redhat.com> <4B71EDDE.6070209@garzik.org> <20100209172157.509ffdb9@redhat.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:sender:message-id:date:from :user-agent:mime-version:to:cc:subject:references:in-reply-to :content-type:content-transfer-encoding; bh=2mS9++ZM2V0VDSRsFhQEcHLNPS64J4NSz7NzqQ0o3cM=; b=C4gS3ynGoeJLtBGnugH481JEHvSjUmsQtWHsXW5Zj75VAz2PYqj49TP5WGKfi9NtPr 5ZRkRakrDCZGF45VTGYiSmP5jGOZPKx/GSFmB9C0HDz3esf7m6BlI8Rez51dWD46zFXg ZFdhMC8ZoQUBBrD//Z/Cad/6+RA9HaFL9D+PA= In-Reply-To: <20100209172157.509ffdb9@redhat.com> Sender: hail-devel-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: Pete Zaitcev Cc: Colin McCabe , Project Hail List On 02/09/2010 07:21 PM, Pete Zaitcev wrote: > On Tue, 09 Feb 2010 18:21:02 -0500 > Jeff Garzik wrote: > >> So, the programs themselves have it right -- debug/verbose is an integer >> not a bool. The internal libraries get that wrong sometimes, though. >> >> Having separate knobs for 'debug' and 'verbose' will be more confusing >> than a single control, IMO. > > I dunno, I think bigger packages tend to drift towards bitmasks. > But I don't have a strong opinion. I just wanted to restore the > per-packet output that you wanted. If you think that an integer is > better, that's fine. The newer code should is more regular and easier > to mold in any desired way. Integer or bitmask, either one is fine with me. Anything but two separate variables, one for 'debug' and one for 'verbose'. Programmers tend to like bitmasks, and users tend to -not- like them if forced to perform binary math. > Actually, I think the whole idea of debug levels is flawed. > We should work on adding a CLD protocol dissector to Wireshark > instead, now that with Colin's help we've settled to a protocol. Long term, I agree. Creating a wireshark patch is definitely a to-do item. For now, while a wireshark patch would be welcome, we are still changing the network protocol. In particular, replication will _definitely_ change the protocol. I was waiting on the XDR patch before making any more protocol changes. That means we would either need to have locally patches wireshark packages, or maintain an out-of-tree wireshark patch in rawhide for a while, until the protocol settles down. Jeff