Re: Future of ident(1)
Has anyone considered that ident(1) works on all the binaries produced by FreeBSD for the first 25 years it existed? Ie, you can run ident on a FreeBSD 1.0 binary... I also believe it can be run on binaries from many other systems. So though we have lost the data in FreeBSD 14, the other data still exists, and though some might call this obscure, others of us find this data pricessless in figuring out from which a binary was made. Regards, Rod > Just in case someone thinks this is something useful, I've got a PR filed > against git to implement custom $Id$. There are no technical issues with > the change but Git developers are not very keen on merging it in, since > they see this as an obscure feature, no one needs. :-/ So if anyone wants > to chime in on GitHub or git mailing list please do so. > > https://github.com/git/git/pull/1074 > > Thanks! > > -Max > > On Fri, Oct 22, 2021 at 2:30 PM Steve Kargl < > s...@troutmask.apl.washington.edu> wrote: > > > All, > > > > With the new world order, what is the future of ident(1)? > > Should it be removed from base? Given a compiled binary > > in base, how does one find the equivalent git info that > > ident(1) used to perform? > > > > Having a few minutes to dust off old patchs for libm, should > > I remove $FreeBSD$ tags in files I touch? For new files, is > > it expected that useless $FreeBSD$ tags should be added? > > > > -- > > Steve > > > > -- Rod Grimes rgri...@freebsd.org
Re: Future of ident(1)
On Mon, Oct 25, 2021 at 09:47:18AM -0700, Maxim Sobolev wrote: > Just in case someone thinks this is something useful, I've got a PR filed > against git to implement custom $Id$. There are no technical issues with > the change but Git developers are not very keen on merging it in, since > they see this as an obscure feature, no one needs. :-/ So if anyone wants > to chime in on GitHub or git mailing list please do so. > > https://github.com/git/git/pull/1074 Or temporary put one file into devel/git/files? > Thanks! > > -Max > > On Fri, Oct 22, 2021 at 2:30 PM Steve Kargl < > s...@troutmask.apl.washington.edu> wrote: > > > All, > > > > With the new world order, what is the future of ident(1)? > > Should it be removed from base? Given a compiled binary > > in base, how does one find the equivalent git info that > > ident(1) used to perform? > > > > Having a few minutes to dust off old patchs for libm, should > > I remove $FreeBSD$ tags in files I touch? For new files, is > > it expected that useless $FreeBSD$ tags should be added? > > > > -- > > Steve > > > >
Re: Future of ident(1)
Just in case someone thinks this is something useful, I've got a PR filed against git to implement custom $Id$. There are no technical issues with the change but Git developers are not very keen on merging it in, since they see this as an obscure feature, no one needs. :-/ So if anyone wants to chime in on GitHub or git mailing list please do so. https://github.com/git/git/pull/1074 Thanks! -Max On Fri, Oct 22, 2021 at 2:30 PM Steve Kargl < s...@troutmask.apl.washington.edu> wrote: > All, > > With the new world order, what is the future of ident(1)? > Should it be removed from base? Given a compiled binary > in base, how does one find the equivalent git info that > ident(1) used to perform? > > Having a few minutes to dust off old patchs for libm, should > I remove $FreeBSD$ tags in files I touch? For new files, is > it expected that useless $FreeBSD$ tags should be added? > > -- > Steve > >
Re: Future of ident(1)
On Fri, Oct 22, 2021 at 03:51:15PM -0600, Warner Losh wrote: > On Fri, Oct 22, 2021 at 3:30 PM Steve Kargl < > s...@troutmask.apl.washington.edu> wrote: > > > All, > > > > With the new world order, what is the future of ident(1)? > > Should it be removed from base? Given a compiled binary > > in base, how does one find the equivalent git info that > > ident(1) used to perform? > > > > It is useful for some rare cases, but in general those cases likely > can likely be handled by readelf / elfdump and adding build info > to the files. I tendto think it's time to retire ident(1), but I'm in no > hurry. > I played a bit with objdump. It seems currently a compiled object and/or binary does not include a git hash. I suppose this makes it only a bit more inconvenient to explore history. > > > > Having a few minutes to dust off old patchs for libm, should > > I remove $FreeBSD$ tags in files I touch? > > No. The current rule is that we're not removing $FreeBSD$ as we touch files. > The plan is to remove it wholesale in the future. This gives maximum > flexibility because stable/12 is still built out of svn and you never > know what might get MFC'd there. > OK. > > For new files, is > > it expected that useless $FreeBSD$ tags should be added? > > > > If you plan to or expet to merge the new file to stable/12, then yes. > Otherwise omit $FreeBSD$. Most new files won't likely be merged to 12 > and even if you make an error in judgement and a new file without > $FreeBSD$ is merged to 12, the impact is going to be minimal. I have no intentions of merging anything. I'll submit re-open a PR and attach the updated to it. -- Steve
Re: Future of ident(1)
On Fri, Oct 22, 2021 at 3:30 PM Steve Kargl < s...@troutmask.apl.washington.edu> wrote: > All, > > With the new world order, what is the future of ident(1)? > Should it be removed from base? Given a compiled binary > in base, how does one find the equivalent git info that > ident(1) used to perform? > It is useful for some rare cases, but in general those cases likely can likely be handled by readelf / elfdump and adding build info to the files. I tendto think it's time to retire ident(1), but I'm in no hurry. > > Having a few minutes to dust off old patchs for libm, should > I remove $FreeBSD$ tags in files I touch? No. The current rule is that we're not removing $FreeBSD$ as we touch files. The plan is to remove it wholesale in the future. This gives maximum flexibility because stable/12 is still built out of svn and you never know what might get MFC'd there. > For new files, is > it expected that useless $FreeBSD$ tags should be added? > If you plan to or expet to merge the new file to stable/12, then yes. Otherwise omit $FreeBSD$. Most new files won't likely be merged to 12 and even if you make an error in judgement and a new file without $FreeBSD$ is merged to 12, the impact is going to be minimal. Warner
Future of ident(1)
All, With the new world order, what is the future of ident(1)? Should it be removed from base? Given a compiled binary in base, how does one find the equivalent git info that ident(1) used to perform? Having a few minutes to dust off old patchs for libm, should I remove $FreeBSD$ tags in files I touch? For new files, is it expected that useless $FreeBSD$ tags should be added? -- Steve