Wow. That's a comprehensive report. It really makes clear what is necessary. I'll definitely review all the places you mention for LFS-7.5. I'll also forward the BLFS only issues to either the blfs-dev list or create appropriate blfs tickets.
Many thanks. -- Bruce Note: Top posting because this is the exception that proves the rule. The original context below is important. akhiezer wrote: >>>> We have been using http://www.pathname.com/fhs/pub/fhs-2.3.html >>>> >> >> >> OK; am having a look at this just now - doesn't look like very many changes >> in the fhs docs; init report &c to follow. >> > > > Here's that report, below. > rgds, > akh > > ------------------------------ Start of Report ------------------------------ > > > Possible Use by B/LFS, of FHS Changes from FHS Version 2.3 to Most-Recent FHS > ('FHS 3.0 Draft 1'). > > 20140210 (10/feb/2014) > > > #### > #========= > # Overview: > #========= > > Most-recent formal release of FHS is version 2.3, released on 29/jan/2004 > (20040129). > > In 2011, work was begun towards a new version. A draft was released on > 16/aug/2011 (20110816), called 'FHS 3.0 Draft 1' . There appears to be no > further commits/revisions to the project after that date, other than two > commits on 20/jan/2012 (20120120) concerning generating txt/html/pdf versions > of the document from the xml source. > > B/LFS currently uses FHS version 2.3 as its most-recent FHS reference. It's > now wanted to see if B/LFS wants to use any of the changes made between FHS > version 2.3, and version 'FHS 3.0 Draft 1' . > > > #### > #===================== > # FHS Old/New Websites: > #===================== > > > Previous home of FHS: > > http://www.pathname.com/fhs/ > > > Current main page for fhs appears to be: > > http://www.linuxfoundation.org/collaborate/workgroups/lsb/fhs > > But it doesn't link directly to the v3.0-draft1 work, and has a broken link to > 'version control repository', and so on; see below for proper link, and some > notes re messiness. > > The FHS material at its 'new-home' at linuxfoundation.org , is a bit messy - > broken links, mazes, (wiki), statements with implied dates attached (but not > the date that one might reasonably expect), slightly different versions of > what is billed in separate places as apparently the same (web-)page, and so > on; > the sort of thing that one commonly sees in long-ish stalled projects &/or > from sizeable 'organisations'. > > > Version 2.3 can be obtained from either of the aforenoted old-/new- FHS homes: > -- > http://refspecs.linuxfoundation.org/fhs.shtml > > http://www.pathname.com/fhs/ > -- > > > Version '3.0 Draft 1' can be obtained via: > > http://www.linuxfoundation.org/collaborate/workgroups/lsb/fhs-30-draft-1 > > On that page, the link, > > "You can see it here: > http://dev.linuxfoundation.org/~licquia/fhs-3.0-drafts/"; > > , is just to a subset of derived formats (txt, html, &c). > > The source is via the next link, > > "Source code is available from our Bazaar repository here: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec"; > > > NB to treat the rendered formats (txt/html/pdf/...) with some caution - e.g. > those obtained via: > -- > (ver. '3.0 Draft 1'): http://www.linuxbase.org/betaspecs/fhs/ > > (ver. 2.3): http://www.pathname.com/fhs/ > > (ver. 2.3): http://refspecs.linuxfoundation.org/fhs.shtml > -- > For, they often differ from the supposedly equivalent files generated from src > Makefiles. Layouts differ (e.g. indentation, flowing of text), (pseudo-)random > ID codes get generated and embedded throughout, different 'bullet' symbols get > used for list items, utf-8 may be used instead of ascii, or vice versa, and so > on. All of which make doing diffs &c really quite ... 'awkward'. Also, the > '3.0 Draft 1' txt/html items, in the docs themselves and in their website > download areas, don't give any reasonably upfront indication - e.g. via > version > numbers, 'release'/revision dates, changelogs, copyright/licence notices - > that > they are version '3.0 Draft 1'; and instead look just like the ver 2.3 docs in > those respects. > > > Given the above-noted messiness, we were careful to cross-check things where > possible, and not take things for granted. As is often the case in such > situations, things become a bit clearer with the src. > > > #========================== > # FHS '3.0 Draft 1' Src/RCS: > #========================== > > > As said, the '3.0 Draft 1' project's rcs is at: > > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec > > > The src for '3.0 Draft 1' appears to be in good order. A main exception is > that the 'ChangeLog' is still more-or-less the same as the 2.3 version; i.e. > it doesn't include entries for the full set of revisions since 2.3 . Some > projects do only update the 'ChangeLog' file at release time; and during > development just use the rcs comments as a working changelog. But here, there > _are_ some early changes made to the ChangeLog file; but none beyond that, > and certainly not covering the full set of revisions since 2.3 . > > > The index-lists of individual commits (newest first, 20 per page) are at: > -- > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/changes/51?start_revid=51 > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/changes/31?start_revid=51 > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/changes/11?start_revid=51 > -- > Currently there are 51 commits. > > > Each commit has its own page with two tabs - where 'n' is the commit/revision > number: > == > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/${n} > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/files/${n} > == > NB that, in the usual way, these individual commit pages contain, inter alia, > (at 1-/2- steps' remove) links to the full contents of the changed files, so > that the user can see the changes in more context than just the diff. > > > For each commit, the diff between the new and preceding revision, can be > downloaded via: > > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/diff/${n} > > , where currently 'n' runs from 2 to 51 ; for n=1 you get essentially a 'null' > diff - the 'n=1' commit is just the import of fhs-v2.3 . If downloaded via > wget, then the local filename is just "${n}" . If the download is done > manually via the 'download diff' link on an individual revision's page, then > the downloaded filename is "${n}_${n-1}.diff" . However, the two versions of > the file, namely the manual- & automated- download versions, are identical, > at least for the few tested here. > > > As one of our cross-checks, we applied the patches in sequence to v2.3 src, > and compared the result with '3.0 Draft 1' src: it matched exactly. > > > We also used the '3.0 Draft 1' Makefile to try to generate the full set of > txt/html/pdf formats: it worked straightforwardly-OK, and the results look > OK. We then compared the outputted files with those available via > 'http://www.linuxbase.org/betaspecs/fhs/' : the latter doesn't have pdf, but > does have txt (utf-8), html (all-on-one-page), and html-chunked (a dirtree of > html docs, approx one per section/chapter in doc - usual stuff). As noted, > the diffs are slightly problematic due to 'behind-the-scenes' formatting > data: but on stripping that out, the essential content of the two sets of > docs - excluding the pdf format, that we don't have an upstream version to > compare against - matched exactly. So, we'd say that the '3.0 Draft 1' > src + Makefile work OK together. > > > #### > #=============================== > # List of Revisions and Comments: > #=============================== > > > The following is a list of the revisions, with for each revision a brief note > on whether it's likely to affect b/lfs or not. We don't specify exactly which > pages in b/lfs would perhaps need changing, as that is best left to those > active developers who have the up-to-speed knowledge of the (fine(r)) details > of how b/lfs interacts with fhs. > > The notes often paraphrase and summarise the revisions: this of course takes > the revisions out of context; we have tried to retain and not distort said > context, and not be misleading - but do bear in mind that they are > out-of-context summaries; the upstream revisions and src, give the > (more-)correct and full(er) context. > > Also note that we are here, for the purposes of this report, relating opinions > and decisions of the FHS authors/contributors/&c as-is, hopefully objectively, > and without adding our own more-subjective commentary: we are not necessarily > advocating or agreeing with the fhs decisions, contents, &c. > > > The list is in order of most-recent revision (i.e. highest-numbered) first. > > > ---- > Key: > -- > line 1: upstream Revision number. > line 2: upstream Revision summary-comment. > line 3: upstream Revision date (yyyy-mm-dd). > line 4: local (akh/cz) date and summary-comment. > line 5 et seq (if any): local (akh/cz) further comments. These 'further > comments' sometimes include, for convenience, a direct link to the upstream > revision's own web page. > ---- > > ---- > 51 > Make xmlto args work with both old *and* new versions. > 2012-01-20 > ##cz20140209## OK; no action reqd for b/lfs. > Is just re Makefile for rendering FHS xml-src in other formats > (txt/html/pdf/&c). > > ---- > 50 > Make xmlto args compatible with older xmlto. > 2012-01-20 > ##cz20140209## OK; no action reqd for b/lfs. > Is just re Makefile for rendering FHS xml-src in other formats > (txt/html/pdf/&c). > > ---- > 49 > Add /usr/share/ppd. (bug 805) > 2011-08-16 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/49 > Adds optional /usr/share/ppd dir(tree) for printer description ('PostScript > Printer Definition') files. > > ---- > 48 > Simplify boot requirements. (bug 801) > 2011-08-16 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/48 > Some changes of emphasis on what goes where re boot files. > > ---- > 47 > Remove reference to window manager rc files in /etc/X11. > 2011-08-16 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/47 > Removed two recommendations: one about naming & locating '*wmrc' files under > /etc/X11 ; and the other about naming window-manager subdirs under /etc/X11 . > > ---- > 46 > Remove /usr/bin/mh. > 2011-08-16 > ##cz20140209## OK; no action reqd for b/lfs. > I.e. 'mh' as in 'MH mail handling system (optional)'. > > ---- > 45 > Fix little wording issue pointed out by Mats. > 2011-08-16 > ##cz20140209## OK; no action reqd for b/lfs. > > ---- > 44 > Add /usr/libexec. (bug 101) > 2011-08-16 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/44 > Formally introduces /usr/libexec in FHS, as optional; acknowledges earlier > common usage outwith FHS; and doesn't deprecate continued use of /usr/lib > for libexec-type material. > Therefore, guess that b/lfs likely has this covered. > > ---- > 43 > More POSIX shell tweaks, plus a fix to a validity bug. > 2011-08-15 > ##cz20140209## OK; no action reqd for b/lfs. > > ---- > 42 > Tweaks to Karl's POSIX shell patch. > 2011-08-15 > ##cz20140209## OK; no action reqd for b/lfs. > > ---- > 41 > Fix Bourne shell references to refer to POSIX instead. > 2011-08-15 > ##cz20140209## OK; no action reqd for b/lfs. > Just changes to more-generic terminology: 'Bourne shell' to 'POSIX compatible > shell'; and likewise the 'bash' and 'C-shell' examples get replaced with > generic-worded versions. > Note that this substituting has left the now-incorrect wording "The default > system and user shells interpreters are already [...]": that gets corrected > in revision 42. > > ---- > 40 > Clear up status of /usr/lib/sendmail and /usr/sbin/sendmail. > 2011-08-15 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/40 > Probably not needed for b/lfs. Just takes the prev requirement > '/usr/lib/sendmail -> /usr/sbin/sendmail (if the latter exists)', and > generalises it slightly for systems with sendmail-compatible MTA. > > ---- > 39 > Add color management directories (bug 77). > 2011-08-15 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/39 > Prescriptions for dirs '/usr/share/color', '/usr/local/share/color', and > '/var/lib/color' . > > ---- > 38 > Remove lib64 restrictions; not used in practice by many grou... > 2011-07-01 > ##cz20140209## OK; no action reqd for b/lfs. > Just removes some perhaps-overly-prescriptive stuff re lib64/lib32/lib . > > ---- > 37 > Tweaks to /run suggested by Lennart Poettering. > 2011-07-01 > ##cz20140209## OK; no action reqd for b/lfs. > I'd expect that all of the items are already now implemented in b/lfs. > (Note that the text 'it is a major security problem if any user can write in > this directory' is still ambiguous: should really, in the context, read > s'thing > like '... unprivileged user ...' .) > > ---- > 36 > Implement /run, and deprecate /var/run. > 2011-07-01 > ##cz20140209## OK; no action reqd for b/lfs. > I'd expect that all of the items are already now implemented in b/lfs. > (NB that the text doesn't anywhere actually say _why_ shift from /var/run to > /run ; am not asking what those reasons are - just noting that the new fhs > text doesn't say it.) > (NB also that the tortured and wrong "unwritable for unprivileged users (root > or users running daemons)" is corrected in revision 37, above.) > > ---- > 35 > Add reference to XDG and GLib for home directories (bug 788)... > 2011-06-28 > ##cz20140209## OK; no action reqd for b/lfs. > Just adds brief note: "A number of efforts have been made to standardize the > layout of home directories, including the XDG Base Directories specification > [...] and the GLib conventions on user directory contents. [...] ." > > ---- > 34 > Remove obsolete refs to XFree86. Patch from bug 772. > 2011-06-21 > ##cz20140209## OK; no action reqd for b/lfs. > Just a few more pieces of transition from XFree86 to Xorg: 'Xconfig' & > 'XF86Config', -vs- 'xorg.conf' . > > ---- > 33 > Fix validity error in previous patch. > 2011-06-21 > ##cz20140209## OK; no action reqd for b/lfs. > Just xml-src tidying (re project Makefile's 'make valid', > 'xmllint --valid ...', &c). > > ---- > 32 > Get rid of /usr/X11R6 and /usr/bin/X11. Patch from bug 73. > 2011-06-21 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/32 > * Removes specific references to 'X11R6', all from xml-src-file 'usr.xml'; > but leaves a few references in 'var.xml', that are still present in > current (immediate-post-revision-51) xml-src. > * Removes requirement for '/usr/lib/X11' symlink, and implies it should not > be present: but then (~14 lines later) makes an adjustment ( > s|/usr/X11R6/lib/X11|/usr/lib/X11| ) that adds it to an existing comment > re global config files; and that is still present in current > (immediate-post-revision-51) xml-src. > > ---- > 31 > Prefer Unicode for /usr/share/dict; add Unicode examples to ... > 2011-06-16 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/31 > Now recommends unicode instead of non-unicode, for some dict(ionary) and > man-page stuff. > > ---- > 30 > Attempt to handle executables in /etc better. (bug 58) > 2011-06-15 > ##cz20140209## OK; no action reqd for b/lfs. > Just adds a line noting what is meant by 'binaries' re 'no (executable) > binaries in /etc' . > > ---- > 29 > Tweak the rationale for root fs requirements. (bug 56) > 2011-06-15 > ##cz20140209## OK; no action reqd for b/lfs. > Tho', the changed material does slightly reinforce the recommendation to > have (or 'leaning towards having') a 'separate /' partition. > > ---- > 28 > Add section in Linux annex for /sys. (bug 53) > 2011-06-14 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/28 > One-para description of /sys . > > ---- > 27 > Remove details regarding ext2 utils and LILO. (bugs 50 and ... > 2011-06-14 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/27 > Removes specific reference to 'additional' mke2fs & lilo stuff going into > /sbin, and instead just has a generic phrase about 'commands relating to > filesystem maintenance and boot loader management' going into /sbin . > > ---- > 26 > Clarify differences between /srv and /var/lib (bug 42). > 2011-06-14 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/26 > Just reinforces/clarifies that, for 'var/lib'-type of data: /var/lib is for > stuff where the directory and file structure of the data (that is under > /var/lib) is not exposed to ordinary users; /srv is for system-wide stuff > that _is_ exposed to ordinary users; and data that is only of interest to > a specific user should go in that users' home directory. > > ---- > 25 > Add docbook-xsl stylesheet to the project, and use it for PD... > 2011-06-14 > ##cz20140209## OK; no action reqd for b/lfs. > More wrestling with rendering of xml-src &c. > > ---- > 24 > Merge from upstream. > 2011-06-14 > ##cz20140209## OK; no action reqd for b/lfs. > Just makefile stuff to get rendered copies (txt/html/pdf/&c) working better. > > ---- > 23 > Clarify /usr/share/misc/magic (bug 39). > 2011-06-14 > ##cz20140209## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/23 > Two items, one not specific to 'file'/'magic': > * (general): Applications using only a single file under /usr/share, may use > /usr/share/misc dir rather than the usual application-specific subdir under > /usr/share ; and similarly for /usr/local/share , for applications > installed 'locally'. (Maybe re {usb,pci}.ids ?). > * ('file'/'magic'): if 'file' installs some auxiliary stuff under > /usr/share/file dir (as modern variants of 'file' do), then rather than > using the old single-file '/usr/share/misc/magic', one can create optional > compat symlink /usr/share/misc/magic -> /usr/share/file/magic . > > ---- > 22 > Fix dependencies on chunked HTML generation. > 2011-05-26 > ##cz20140209## OK; no action reqd for b/lfs. > > ---- > 21 > Fix PDF generation, and create build documentation. > 2011-05-26 > ##cz20140209## OK; no action reqd for b/lfs. > > ---- > 20 > fix bug 62 (really this time) > 2011-05-20 > ##cz20140208## OK; no action reqd for b/lfs. > Just multiple s/can not/cannot/ . > > ---- > 19 > a litle reorg in informal Attributions tracker > 2011-05-19 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 18 > temporary fix for tag sync problem > 2011-05-09 > ##cz20140208## OK; no action reqd for b/lfs. > Just internal xml/src plumbing. > > ---- > 17 > extend no-subdirs rule to more places (bug 48) > 2011-05-09 > ##cz20140208## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/17 > "There must be no subdirectories in ": added /sbin, /usr/bin, and /usr/sbin > to the list. > > ---- > 16 > grammar change (bug 57) > 2011-05-09 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 15 > change to use removable (bug 102) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > Just multiple 'typo' 'fixes' s/removeable/removable/ . > > ---- > 14 > add tile explanation for /usr/local/share (bug 427) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > Commit comment should have s/tile/title/ : and revision just explains, > "/usr/local/share : Local architecture-independent hierarchy" . > > ---- > 13 > undo requirement for /bin/gunzip as a link (bug 107) > 2011-05-06 > ##cz20140208## Might want to adjust b/lfs. > Ref: > http://bzr.linuxfoundation.org/loggerhead/lsb/devel/fhs-spec/revision/13 > Don't need to have gunzip or zcat, if present, as symbolic or hard links to > gzip. > Bruce (Dubbs) is one of the folks attributed on this revision; so maybe > already done in b/lfs. > > ---- > 12 > clarify required/opt subdirs of /etc (bug 59) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > The change might look a bit messy, or not immediately clear, in xml-source: > but it's just reinforcing that '/etc/opt' is required and that > '/etc/{X11,sgml,xml}' are optional (only required if relevant subsystem is > installed). > > ---- > 11 > remove contributor emails (bug 760) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 10 > typo (bug 327) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 9 > formatting fix (bug 764) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 8 > apply editing fixes for bugs 61, 62, 63, 65, 66 > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > The long-ish list of example files, is the same as for prev-commit. > > ---- > 7 > close out bug 428 > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 6 > update bug links in changelog (mail list submission) > 2011-05-06 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 5 > various noodlings on producing pdf > 2011-05-05 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 4 > update list/contact info; draft.xml isn't part of spec so re... > 2011-05-05 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 3 > polish > 2011-05-05 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 2 > convert to xml > 2011-05-05 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > 1 > Initial commit of FHS 2.3. > 2011-04-19 > ##cz20140208## OK; no action reqd for b/lfs. > > ---- > > > #### > # Fin. > # > ## > > > > ------------------------------ End of Report ------------------------------ > > > > > -- > -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
