655262 - fix scratchbox deployment

Status: RESOLVED INCOMPLETE

(Release Engineering :: General, defect, P5)

Description

[Dustin J. Mitchell [:dustin] (he/him)]

Scratchbox is currently installed via two rpms, which seem to be wrappers around untarring a tarball.  This has a few problems.  The RPM must be downloaded in its entirety before being installed, which doubles the (signficant!) space required.  It's often impossible to install scratchbox on a VM that is not freshly imaged.

We also see repeated re-installs of scratchboxes.  I'm not sure why, but I suspect it has something to do with the Exec[check-scratchbox-install] resource.

Comment 1

[Aki Sasaki (not active)]

We should weigh the downsides of not fixing this against the potential of de-supporting Maemo in the foreseeable future.

Comment 2

[Dustin J. Mitchell [:dustin] (he/him)]

It's one of the more difficult forms of slave-wedged to get out of, and also chews bandwidth until it's fixed (since it downloads scratchbox almost continuously, with 60 second pauses between each download).  So I'd like to fix it.

Comment 3

[Chris Cooper [:coop] (he/him)]

(In reply to comment #1)
> We should weigh the downsides of not fixing this against the potential of
> de-supporting Maemo in the foreseeable future.

Aki: any idea when that might be? Is it tied to a particular version?

Comment 4

[John Ford [:jhford] CET/CEST Berlin Time]

if we do decide to fix it, we need to be very very careful to capture the pre(un)/post(un) logic in the rpm.  Without the special steps, we could easily screw up the entire system.

How often do we have to reinstall?