The install documentation says to skip extracting the portage snapshot if a network installation is being done. I would strongly recommend that this is changed to an recommended/optional step even in the case that a network install is being done. Furthermore, it should be recommended that a portage snapshot is downloaded and extracted should it not be available on the installation media. The user can then continue on to "emerge sync" etc. This bug is filed as a result of discussions on #gentoo. It seems that people there who know what they are doing, do this anyway in order to save time - the initial emerge sync is a lot quicker this way. Reproducible: Always Steps to Reproduce: 1. 2. 3.
I think this would lead into confusion for GRP users which snapshot they should extract. So i would rather suggest to advise our users to unpack the snapshot provided on the livecd (if there is a snapshot on the user's particular livecd) and then "emerge --sync" if they don't want to use GRP packages. Any other comments?
Perhaps place it in the tipsntricks document as it's a time-saving trick that isn't mandatory nor important for the installation? I'd like to use the tipsntricks document as a "staging" document for this kind of tips/tricks. If at any time we decide to do a major overhaul of the installation instructions those tips/tricks can be merged in intelligently. Imho, adding it to the installation instructions in their current form would indeed pull more question marks than exclamations :)
I am inclined to think it will make the docs no more complex. It is simply a matter of changing "if you are doing a network install, skip this step" to "optional for network install(1 line explanation of advantage)/mandatory for grp install" with the part after the "/" being what is already in the docs.
I suppose this is the same thing as what happens when using emerge-webrsync. Perhaps we can tell users that emerge-webrsync isn't only for when you're unable to "emerge --sync" but that it can very well shorten the Portage tree download time?
Oh bleh, emerge-webrsync is slower because it uses rsync to update the Portage tree (from a temporary location to the official one). I'm going to implement Tobias' proposal...
The number of devs who need to be told what emerge-webrsync does, is suprisingly high ;-) I suppose very few actually use it. Tobias' suggestion would be better indeed. Actually, IMHO, all users should unpack the tarball from the CD, then run emerge --metadata if networkless (I think) or emerge --sync if connected
what if they use a minimal livecd? download a stage tarball instead?
Ah yes, of course there is no tarball on the minimal CD. I'd say we keep it simple: if Universal CD, unpack archive. if networkless, emerge --metadata else emerge --sync Using a minimal CD means downloading a lot and maybe it doesn't make enough of a difference. FYI, `rsync` when you have no data just took me 10 minutes. How long do you think it would take anyone to find a mirror, find a snapshot, download it, unpackit, run emerge --sync ?
Created attachment 44874 [details, diff] hb-install-stage patch
Created attachment 44875 [details, diff] hb-install-system patch
The above two patches update the installation instruction so that: - each user unpacks a portage snapshot (either LiveCD or downloaded) - each user either runs "emerge --sync" or "emerge --metadata" Could someone verify if these updates are sane? They're a bit too important and I'm a bit tired so I might have missed something important.
neysx was confused until he realized swift had already committed the first patch when doing s/Gb/GB/ Both patches have now been reviewed, edited and applied. I hope we got it right.
Holy malloney, sorry about that. Wasn't my intention of doing this with the size changes :'( Thanks for taking care of it.
