Summary: | documentation and the USE flag | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Mr. Bones. (RETIRED) <mr_bones_> |
Component: | [OLD] Core system | Assignee: | Robert Coie (RETIRED) <rac> |
Status: | RESOLVED FIXED | ||
Severity: | enhancement | CC: | alain, vapier |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: | patch for perl.ebuild |
Description
Mr. Bones. (RETIRED)
2003-02-26 05:14:15 UTC
(1) we will not check for doc USE inside an ebuild because we'd be doing it in just about every single ebuild we have ... yeah, we have over 3000 ebuilds ... (2) doc is for 'extra documentation' ... usually defined as documentation that is generated by extra utilities or something ... this is under discussion its under discussion but what the user asked for originally is not I realize that this discussion is ongoing regarding how to prevent packages from installing documentation globally, however... The documentation Michael mentions in the Perl package is in fact API documentation above and beyond the manual pages. There's a full 13M of html files! It seems to me that this would be an appropriate use of USE=-doc, don't you think? that question should be brought up with mcummings ;) Created attachment 10860 [details, diff]
patch for perl.ebuild
Patch to not install the massive amount of html docs that perl supplies if
use doc is not specified.
Conversation seems to have died out on this issue. While I can appreciate SpanKY's comment that there are a lot of ebuilds and checking `use doc` for each one might seem daunting, I think this is a pretty good ebuild to make an exception for since the size of the files in question is pretty compelling, at least for me. Additionally, since I've supplied a patch the total amount of work is pretty close to zero. Oh, I should also note that the patch also takes out 25 seconds of sleeping that is pretty annoying. Frankly, most people are probably not going to sit and watch perl merge to their systems so the sleep simply adds time rather than value to the proceedings. I did leave the sleep 10 in the case of `use threads` as further punishment to those people who choose to merge perl with thread support as that is sure to be just the beginning of the pain. Fixed in CVS in a silent upgrade to 5.8.0-r10 |