From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Petazzoni Date: Thu, 16 Jan 2020 10:31:20 +0100 Subject: [Buildroot] [PATCH] infra: add the transient download mechanism In-Reply-To: <534f269547ead6d86e7a571978b54d79cd24cd0d.camel@orolia.com> References: <20200115203753.12110-1-yann.morin.1998@free.fr> <534f269547ead6d86e7a571978b54d79cd24cd0d.camel@orolia.com> Message-ID: <20200116103120.7b41295a@windsurf> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Hello, On Thu, 16 Jan 2020 09:11:32 +0000 Nicolas Carrier wrote: > One small remark, to be able to reproduce the build, on top of this > mechanism, there should be: > * a way to store the actual VCS revision used for the build (to at > least know what was just built!) That seems quite easy to do. > * a way to force the build to re-use a particular set of revisions I don't see how that can be done however, and how that would work from a user interface point of view. Sure the first build could store the exact Git commit used for building each transient package, and store that "somewhere", and then re-use that for the next build. But how does the user decide whether what he wants is to re-do the build with the same commits or not ? Where is that "somewhere" ? > That's part of why I don't think that this kind of feature is a good > thing to integrate into buildroot. I prefer to delegate it to tools > meant for that (using a mix of local packages and OVERRIDE). Different people/companies have different work flows. We have people telling us that using local packages and OVERRIDE for their CI is not very convenient/practical. > And like said before, I think it's preferable to have a optional > feature to prevent branches from being used, making the build fail. The feature proposed by Yann *is* optional, as it's conditional on having a specific variable set in the package .mk file. However, one question that is raised by Yann's patch is whether we want to forbid using branches when TRANSIENT_DOWNLOAD is NO. Best regards, Thomas -- Thomas Petazzoni, CTO, Bootlin Embedded Linux and Kernel engineering https://bootlin.com