[COFF] Personal Preference vs Group Productivity (was Re: [TUHS] Re: End of an era: the last ATC (USENIX Annual Technical Conference))

[TUHS to Bcc; Cc'ed to COFF to redirect for follow-ups] On Sun, Jul 20, 2025 at 8:24 AM Christian Hopps <chopps(a)chopps.org> wrote: I think some folks may have interpreted Larry's statements as stating that, if you worked for him, you used the tools he mandated. But I don't think that's what he meant (at least not about text editors), and I think that your interpretation is correct: use whatever works for you, but don't waste a bunch of other people's time pushing the virtunes of your preferred tools on those who already have things that work for them. I think that's fair. That said, there is a qualitative difference between the sort of banter people may bat around over lunch, and someone seriously trying to get everyone else to use their preferred editor (editors are just one example, though interesting in this regard because choosing them seems to be such a deeply personal thing). I think the discussions on these mailing lists are closer to the former than the latter, and as long as it remains friendly, I really don't see any harm in that. But this got me thinking that surely there are decisions made on a project basis, where individual preference by itself is not, and cannot be, the deciding factor for use. Perhaps the most obvious is code-formatting styles, but others may be choice of build tool, revision control system, implementation language, and so on. Where does one draw the line between the tools an individual chooses for their own use, and those mandated by the group? I suspect there is no one, good, universal answer. Take implementation language as an example: "I'll write my part in C, you write yours in FORTRAN..." sounds highly suspect at first, but _might_ make sense if "I'm" tasked with writing the IO part and "you're" taking on the numerical bits and have to interface with a numerical library written in Fortran. Indeed, I've found that when people try to define some kind of general rule for deciding this or that, they tend to be trying to simplify really complex things in a very superficial way that's not reflective of actual practice. But maybe there is some utility in thinking about this generally. Perhaps a reasonable first-order approximation for finding a dividing line might be taken from the answer to the question, "is this something that really only impacts me, or is it going to have an impact on my peers, as well?" Editors sort of make sense here; by and large, as long as I can use the tools effectively, my peers aren't going to be impacted if I use tool A to enter text versus tool B. Similarly with whether I prefer light text on a darker background, or dark text on a lighter background, what font I use, type size, what window manager I use, and so on. But on the other hand, if I demanded that I be able to use My preferred code formatting style regardless of what the rest of the code used, then that _does_ have an impact on those around me, so deserves more thought. And _that's_ the kind of thing that people can argue about endlessly until finally someone just makes a decision. I've said this before, but Google's style guides were a great case in point here. To be honest, I didn't like them and thought that the resulting code was ugly. But after a few weeks, I just stopped noticing that, and I had to admit that they were a great aid in being able to approach a codebase with billions of LOC in it. When `clang-format` came along and took out a lot of the tedium of making your code conform to the style guide, it really freed up a lot of mental capacity to start thinking about real problems, design, and so on, and not just where the braces went. But that was all predicated on someone just Making a Decision. Personally, I've found it very helpful to maintain some level of proficiency in a variety of different text editors, using different tools for different things. Part of this is the whole, "when in Rome, do as Romans do..." sort of thing: I'm not looking for `vi` on VMS or Multics or VM/CMS or something, but I'll also happily use emacs to write Lisp, and on plan 9, I'll use sam or acme for C or troff or whatever. On my Mac, Zed or VSCode work pretty well. On remote Unix machines, a lot of the kids are hip on this "Helix" thing these days. Speaking of plan 9.... One nice thing about a system that embraces true resource sharing is that, when one can construct an environment that contains the set of all the resources one needs to work with and then one can bring one's local toolset to bear on those resources. Plan 9, where resources are represented by files in mutable namespaces, really exemplifies this, which is distinct from a more "remote access" model, where I use something like ssh or mosh or telnet to login to a remote system. In the remote access model, I'm constrained by whatever tools are available on a remote system. So if I'm logging into some remote server over SSH or something, it's handy to know whatever editor may be on the distant end. It's a shame more systems aren't like plan 9 in this way, but they aren't, and for better or worse, I've still got to use them, and it's so much less personally aggravating to care deeply about what tools are available in that context. - Dan C.