On 7/1/26 15:15, Mark Millard wrote: > On 7/1/26 09:17, Alexander Ziaee wrote: >> On 2026-07-01 08:57 -04:00 EDT, "Larry Fahnoe" <[email protected]> wrote: >>> My bringing up my own sysadmin experience on other operating systems did >>> not add meaningfully to the conversation and has needlessly stirred the pot. >> >> I think your words really added meaningfully to the conversation, it brought >> 3 important ideas to light. From my end as a writer I learned that people >> are interpreting the information landscape to mean that FreeBSD has a unique >> issue, when actually VMS and Linux operate the exact same way. > > Long ago I used VAX/VMS and so was familiar with the likes of ;2 file > name suffixes in that context. > > I never thought that the general type of issue is unique to FreeBSD or > to pkg . > >> So I have a lot more context to reflect on needed improvements to our >> information landscape. Second, it shows that actually we have the >> opportunity to solve an unsolved problem in operating systems right now. >> That is particularly exciting to me because I really love it when we have >> one of the oldest continuously maintained codebases in the world and are >> still in places on the bleeding edge working on unsolved problems. Third, I >> think this was informative for many of our readers who administer mixed >> environments (which is probably most of us, including me) who may now know >> that they have and should seek to resolve failed configuration merges in >> their other environments. >> >> Plus, the thread didn't catch fire, so w/rt stirring the pot, its all good. >> >>> I believe this to be a pkg issue, and downstream of that a documentation >>> issue. >> >> I shared the concerns I'm trying to balance, do you have any ideas how we >> can make the docs better? >> Currently, I've documented these in the 15.1 upgrading guide, and the pkg(8) >> manual. > > I do not automatically (re)read the pkg(8) manual to learn about or > remind myself of the specifics of what pkg install and pkg upgrade do or > requires-of-me for aspects that the command output is silent about. So > far as I know, no other pkg commands generate such files.
Actually, pkg add might also be a source of one or more of the 3 types of files. Also, pkg delete might generate *.pkgsave files to preserve the content of files that had some non-default/non-upstream content. So: 4 commands, not 2. > In my view > each of those 2 man pages for should at least explicitly indicate to > where to look up information about the 3 types of files at issue, > including use of notation that allows one to be able to identify the > naming conventions involved. > > But I also think that the command output should explicitly indicate if > the results happen to have generated files to go looking for. That > allows knowing one can avoid looking when no such file was actually > generated --and when one needs to consider looking otherwise (for each > type with a positive count). The man pages can not cover such command > instance information that guides later actions. > >> I don't want to duplicate information because it makes an unsustainable >> maintenance burden, but we can link to these, or if it makes more sense we >> can move them, and we can also link to them. Further, the descriptions >> themselves can always be improved. >> >> Best, >> Alex >> > > -- === Mark Millard marklmi at yahoo.com
