From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <4e9947d7c6d27e8f4f580ac2434ab23e@proxima.alt.za> To: 9fans@9fans.net Date: Mon, 25 Mar 2013 08:27:33 +0200 From: lucio@proxima.alt.za In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Subject: Re: [9fans] gcc not an option for Plan9 Topicbox-Message-UUID: 340dfd62-ead8-11e9-9d60-3106f5b1d025 > Yes, I run Go on native Plan9, Go breaks away from a number of traditions that have long become obsolete and that is its main merit. The price is not only in having to adjust to the change, but also in some sacred cows being slaughtered in the process. But Go also opens the door to better ways of doing things. The build system, raw as it still is, is streets ahead of any conventional build system, but it is tightly coupled to the language. Portability across platforms is much easier, in the Plan 9 tradition, but requires a set of build tools ([568][ac]) that users are not familiar with and [568]l becomes the new bottleneck, to many users' surprise. Cross-development - my favourite feature - becomes much easier, but I am having a great deal of trouble getting my head around all the complications it brings with it. Philosophically, Plan 9 has rattled the proverbial cage and Go is an earthquake by comparison. The outcome is still to be evaluated. But not everyone is going to see it in the same way. Of relevance here is that if Rob and Russ and Ken had let considerations such as pampering slow hardware, we'd have a different language and many features would not be available. At the same time, the need for a slim version of Go will grow with acceptance of the fat model and then people like Kurt may be inspired to restore in the linker the ability to trim libraries of unused modules (don't hold your breath!). If the Go developers had started from the other end, as I would have been tempted to do, the outcome would definitely look nothing like what we have. The nice bit is that there are enough people out there to consider such options and some of them are actually willing to publish their efforts. The people who insists that ONE tool should encompass all these options are those who are too unproductive to do it themselves and fail to see that no-one owes them. In my other life managing a backpackers, I see way too many young people who seem to think that our generation somehow owe them something they are able but not willing to seek for themselves. I could tell you where most of them seem to come from, but I'm sure that would be unfair to all those they leave behind while spending money they did not earn to travel in comfort around the world. ++L PS: Gorka is making amazing progress with the plan9/arm port and the reason I know is that I've just tested his latest efforts on the Sheevaplug and the present obstacle does not seem unsurmountable - but it is very real, so "it's not working yet". Watch golang-dev on Google Groups for updates.