|
From: | Robert T. Short |
Subject: | Re: templated bitwise operators |
Date: | Fri, 25 May 2012 09:24:56 -0700 |
User-agent: | Mozilla/5.0 (X11; Linux i686; rv:12.0) Gecko/20120428 Thunderbird/12.0.1 |
On 05/25/2012 08:58 AM, Michael D. Godfrey wrote:
The following is just a personal opinion and preference. I think we should be very careful with using tools that are too far ahead of their time. In particular, I think we should attempt to use features available with debian stable unless there is a powerful reason not to.On 5/25/12 4:51 PM, Jordi Gutiérrez Hermoso wrote:What's the earliest gcc version we're targetting? Is it still 4.1 or thereabouts? I can't wait to use C++11 features. I'm hoping next year we can start doing that. - Jordi G. H.I think that as long as Redhat Enterprise 5 is in fairly wide use 4.2.1 will still need support. RHEL 6 is the current release, but many users find it a big hurdle to switch. Redhat does not advise attempting an upgrade. Only a fresh install is reliable. This does not help the migration. Also, in an environment where there are multiple systems mixing 5 and 6 seems to cause problems. Sad:-( mdg
The reason I say this is that when a feature is used from, say, debian testing a whole series of general stability problems arise totally unrelated to octave. Since I depend on my machine to make a living I really begrudge time spent tracking down problems that are introduced by a buggy version of some software or by changes introduced in a package that I am familiar with. Updating to new versions of debian at release time is plenty often enough to deal with these problems.
That said, I don't think we should make this a strict *rule*, but a general guideline.
Just my opinion. Bob
[Prev in Thread] | Current Thread | [Next in Thread] |