After seeing much discussion in the forums, and since I myself is on a modem, I'd like to offer the following solution to the 'batch fetching' problem. 1) Have an online html (or text) page which lists all SRC_URIs, which is taken from all ebuilds in all categories. (I.e. traversing the entire portage tree.) This would enable users who does not yet have Portage to take that file and input it to wget, or similar, to download the up-to-date distfiles. This should of course be integrated with the on-line package database, allowing users to see all SRC_URIs in the package details. 2) Have the option of inputing package names (ebuilds) to emerge using stdin. Linux has a 128 kB command line limit, if I'm not misstaken, and this probably means I'm not able to put _all_ packages on the emerge command line, to do something like 'emerge --fetchonly packages...'. It also makes scripts look nicer (without nesting backticks...) 3) Have emerge output the list of _all_ SRC_URIs (including dependencies) for a particular command line. This would be like --fetchonly, except even less is actually done. ;) 4) Allow for wildcards in the package name specification of 'emerge'. I tried doing 'emerge --fetchonly net-mail/.+', but this fails, so I assume regexp support does not exist except when searching. The first would not be hard to implement when the third has been implemented, all that is needed then would be the profiles/categories and the 'ls' command. If number four is implemented, neither profiles/categories nor ls would be needed to implement 1). Reproducible: Always Steps to Reproduce:
emerge -efp packages if you ever need more than 2-3 thousand entries on the command line, there is a major problem. Most system use less than 350 explicit packages. The rest are dependancies. I have 200 on my system.
*** This bug has been marked as a duplicate of 1661 ***