pledge(2) & snapshots of -current

[ocicat]

FYI. I'm passing on the following to those who run -current snapshots.

http://marc.info/?t=146821964900001&r=1&w=2

pledge(2) is a great addition to the OpenBSD 6.0 landscape, but there is fallout users need to be educated upon. The ports@ thread above describes how pledge(2) is being adjusted here in the latter part of the OpenBSD 6.0 development cycle.

I mention this here as this also raises the question about the purpose of snapshots in the OpenBSD world. Snapshots are a means for developers to share amongst themselves the current status of code checked into CVS freeing developers from the need to be constantly recompiling code themselves. The fact that snapshots are available to the world doesn't change their fundamental purpose.

Snapshots are not simply a recompilation of code checked into the "development branch" of the project's CVS repository. Snapshots may also include code which has not been checked into CVS as it is deemed experimental. Developers (& Theo...) may want code to receive a wider testing audience before checking in these new features into CVS. Augmenting snapshots to include uncommitted code is the means used to expose new work to a larger test pool.

Uncommitted code also may be added to snapshots for an extended period too. How the decision is made as to what to include & for how long is a decision made by the developers. The general OpenBSD community (meaning everyone else who are not developers...) doesn't have any say into this practice, nor should we.

There is an alternative for those running -current who don't want to be running uncommitted code -- simply recompile the kernel & userland as described in release(8).

[ocicat]

Interestingly, the www/chromium port has now been corrected in CVS:

http://marc.info/?l=openbsd-ports-cv...0356603668&w=2

However, users of -current will need to wait for the next bulk build for the Chromium package to pick up this change.