Are you the author of -m? If not, maybe the author of -m should knows how it works with -s? If not, maybe this should be documented anyway?
On Mon, May 11, 2020 at 5:01 PM Eric Blake <ebl...@redhat.com> wrote: > On 5/11/20 4:18 PM, Peng Yu wrote: > > I used real files (already sorted) to test whether having -s or not > > affect -m. But I have not made minimal example input files so that is > > why I am not sure about my conclusion. > > > > But the command to try is basically `sort -m -k sort_fields files...` > > or `sort -s -m -k sort_fields files..`. > > That's closer - it shows a pseudo-command line you attempted. But it > still does not lend itself to reproducibility, because we don't know > what 'sort_fields' you used, nor what 'files..' contain. > > You also didn't state whether you tried the --debug option, to see if > the presence or absence of -s showed enough debugging crumbs to prove > that you at least tried to analyze the problem yourself. Nor did you > mention whether you read the source code (it _is_ open source, after > all, so instead of asking someone else to do your homework, _you_ can > find the answer). > > > > > I assume to authors who made -m and -s. My question should be clear? > > Unfortunately, your assumption is wrong. A clear question is one that > includes actual examples, and not one that forces someone to reproduce > the work that you could have already provided them. Put it this way: > suppose it took you 5 minutes to come up with a test case, and that > there are 100 list readers interested in your problem, each of whom then > take another 5 minutes to reproduce the setup from your vague > description. Then you have cost 505 minutes of collective time; and > your original work plus the work of each reader results in a very low > signal-to-noise ratio (5/505 is less than 1% new discoveries, and more > than 99% rehash). But if it takes you an additional 5 minutes to polish > your query into an email that can then be copy-pasted into a terminal so > that each reader can reproduce the problem in 5 seconds, then your > initial 10 minutes of effort (which is indeed twice the work on your > part) plus 500 seconds of list readers' time results in a much better > ratio of useful new work (5/18.3 is > 27%). Although it costs you more, > your efforts to make everyone else's life easier is magnified by the > number of readers benefitted by your extra efforts. (And that's why I'm > spending so long in writing my reply - to try to teach you that your > historical style of questioning leaves a lot to be desired, as well as a > possibly-futile attempt on my part to get you to recognize that the more > effort YOU put into a good bug report, the less likely you are to be > habitually ignored as someone who merely wastes time). > > -- > Eric Blake, Principal Software Engineer > Red Hat, Inc. +1-919-301-3226 > Virtualization: qemu.org | libvirt.org > > > -- Regards, Peng
