Announcement

Collapse
No announcement yet.

Microsoft Announces "Windows Terminal" As Linux-esque Inspired Terminal For Windows 10

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    Originally posted by nils_ View Post
    The problem is that some corporations have a corporate IT department and some odd security policies where they want to manage developer workstations through all the tools that are provided by Microsoft / Windows, forcing developers to run Windows (sometimes even without local admin access). Things like WSL and Docker on Windows allow them to continue this misguided policy, if anyone complains about not being able to run their favourite tools they can just point them to WSL.
    Can't speak to others, but in my office, yes developers are required to have a standard corporate desktop i.e. windows with outlook and office so they can participate in the standard corporate services. But then to perform their development work, they have a separate 'developer pc' where they define what OS, software, tools, etc. they require to do their jobs. Most of our devs run RHEL Workstation on their development pc, as that aligns with our server instances.

    It makes sense that corporate can dictate what is required to participate in corporate IT services. But it makes no sense whatsoever for corporate to dictate what a developer needs to do his/her development work. That's like going to a restaurant, and telling the chef which knives, cutting boards, and pots he should use to cook your meal.

    Comment


    • #42
      Originally posted by torsionbar28 View Post
      Can't speak to others, but in my office, yes developers are required to have a standard corporate desktop i.e. windows with outlook and office so they can participate in the standard corporate services. But then to perform their development work, they have a separate 'developer pc' where they define what OS, software, tools, etc. they require to do their jobs. Most of our devs run RHEL Workstation on their development pc, as that aligns with our server instances.
      That does sound a bit cumbersome, then again you can't really use Outlook for any serious e-Mail communication anyways... Though I have to say the browser based stuff from MSFT works fairly well.

      It makes sense that corporate can dictate what is required to participate in corporate IT services. But it makes no sense whatsoever for corporate to dictate what a developer needs to do his/her development work. That's like going to a restaurant, and telling the chef which knives, cutting boards, and pots he should use to cook your meal.
      Love the analogy.

      Comment


      • #43
        Originally posted by torsionbar28 View Post
        Can't speak to others, but in my office, yes developers are required to have a standard corporate desktop i.e. windows with outlook and office so they can participate in the standard corporate services. But then to perform their development work, they have a separate 'developer pc' where they define what OS, software, tools, etc. they require to do their jobs. Most of our devs run RHEL Workstation on their development pc, as that aligns with our server instances.
        In my office we do the same. You don't work upstairs do you? Ha. But no, most of our people develop on RHEL6 systems. With a specific java version and eclipse. Other than those 2, pretty much everything else has to be available on the standard RHEL6 image, no 'extras' repository or anything. Had someone needing ImageMagick the other day and they were freaking out about trying to get a license and stuff. I told 'em it's include on the iso, and everyone was like "phew, we don't have to worry then". Rereading your statement, I really do wonder if you work upstairs now.

        Comment


        • #44
          Thanks M$FT but I'll just keep using mobaxterm anyhow. Cygwin > WSL.

          And I for one DON'T want Microsoft (Or IBM or Oracle) to "just do Linux" Keep them far away from it. They are turning Linux into an application layer to run on their own OS's.

          Comment


          • #45
            Originally posted by k1e0x View Post
            They are turning Linux into an application layer to run on their own OS's.
            For better or worse, this is exactly what's happening. And, like it or not, it's a totally sound business decision, you have to give them credit for that.

            Comment


            • #46
              MS is adopting more and more *nix features which is a good thing. First they added ssh (only compatible with newer distributions, so you still need putty). After that they added curl (but I really miss wget - the 3rd windows party binaries crash for me) and now a new Terminal with better Linux support and even a Linux kernel for WSL2. Something that is still too slow and breaking about 1-2 times a year Antivirus or other applications are updates. Also the default way to install 3rd party application updates not managed by Steam or other integrated update services is time consuming and some online update routines for apps/bios are introducing more security issues than they solve. The Windows Store is too limited, waiting for a better solution...

              Comment


              • #47
                Originally posted by enzo1982 View Post
                Well, it's easy to verify. Time something like this:
                Code:
                #!/bin/bash
                for i in {1..100}
                do
                cat Readme
                done
                Here are my results with a 10kB Readme file:
                Ubuntu 18.10: console output: 0.64s / > /dev/null: 0.11s
                Windows 10: console output: 10.40s / > /dev/null: 2.45s

                So, the difference between console output and no output on Windows is 8s for ~100kB while Linux does that in 0.5s.

                A Microsoft guy wrote on HackerNews[1] that the new Terminal would be ~10 times faster wrt to output thanks to accelerated text rendering. This would put it in the same ballpark as a Linux console.

                Of course I could always suppress output and be fast with the current console too. But I don't want! I want to see what's going on when I'm running something.

                [1] https://news.ycombinator.com/item?id=19841019
                I just tested the same.

                real 0m0.135s
                user 0m0.065s
                sys 0m0.035s

                I don't think this is a bottleneck. I'm surprised it could be, and a bit shocked. But that's "real" time which is not system time. The output is delayed, but the software ran in the same time. I have to do more tests.

                Comment


                • #48
                  Originally posted by c117152 View Post

                  This was explained in the context of java but it's valid here as well: https://stackoverflow.com/questions/...on-performance

                  Basically any I/O, printouts included, hurts performance. At the lowest levels, the context switching trying to schedule different operations alone is very costly since there different physical threads involved. You can also look at the different cache lanes and find thousands of cycles being wasted as each println is being evaluated and de-prioritized.
                  Maybe. But that does not mean it's not wasted otherwise. Using hugetables (largepages) might remove more of that waste than increasing fps output. Which I still am not convinced drives the show.

                  Comment


                  • #49
                    Originally posted by torsionbar28 View Post
                    Can't speak to others, but in my office, yes developers are required to have a standard corporate desktop i.e. windows with outlook and office so they can participate in the standard corporate services. But then to perform their development work, they have a separate 'developer pc' where they define what OS, software, tools, etc. they require to do their jobs. Most of our devs run RHEL Workstation on their development pc, as that aligns with our server instances.

                    It makes sense that corporate can dictate what is required to participate in corporate IT services. But it makes no sense whatsoever for corporate to dictate what a developer needs to do his/her development work. That's like going to a restaurant, and telling the chef which knives, cutting boards, and pots he should use to cook your meal.
                    Briefly worked at a place a few years back, and you had to get the admin to give you privileges to install a bridged VM, and then once you had a bridged VM, you then had full control over it. All the developers were working inside a VM. Often corporate IT admin policy doesn't really cater for developers needs, and it's a fight to get (and wait for) your admin to open up ports, reserve IP addresses etc..

                    Comment

                    Working...
                    X