jam@elrond.CalComp.COM (Julie A. Melbin) (04/14/88)
I know someone can help me out there :)
Upper management needs convincing that software engineers (sic) are
more productive using a multiple window windowing system over a vt100 type
(cheap) system. We all know the truth, but vt100 clones are cheaper that even
the cheapest (uh, sorry - least expensive) multi window system - such as X, and
so therefore, those that control the money think that cheap is better. If I had
some hard proof of how much more productive we could all be, each with our own
X system (diskless is fine), we might even stand a chance of getting a few in
house for development work - yippie!
Thanks forever
Email ...{harvard, decvax}!elrond!jam or jam@elrond.CalComp.COM
or 603-885-8139 daysjam@elrond.CalComp.COM (Julie A. Melbin) (04/17/88)
A bit okay I posted a request for data to try to convince those with the
money why multi window systems are a more productive tool for engineers
than simple (old) ascii terminals. I want to thank those that have responded,
and am still looking for whatever had data/insight you want to share. Someday
we will win this battle!
...{harvard, decvax}!elrond!jam or jam@elrond.CalComp.COMjam@spray.CalComp.COM (Julie A. Melbin) (05/12/88)
In summary from the folks on the net: Simple scenario: three tiled windows, big one running screen editor like emacs, you edit and then write the file (but don't quit emacs); second is command shell to run compiler which puts errors in file; third is where you more the error file. Never need to print listings and edit,compile, then debug is very fast. Add fourth window to view include files to get the calls correct. However, if this is not obvious to your managment, you should find another job ASAP, since the productivity and probably the products are definitely suffering from brain damaged management. A b/w Sun 3 or microVax per programmer is the minimum requirement. Anything less is a sweat shop or technically obsolete. Randall Neff@sierra.stanford.edu From harvard!SEI.CMU.EDU!sr Thu Apr 14 14:47:23 1988 Gentlemen - The acknowledged expert on empirical studies of programmers is Prof. Ben Shneiderman at the Univ of Maryland (e-mail ben@mimsy.umd.edu). Why not ask him? You can tell him that i sent you. - Stan From ucbvax!ucsd!sdsu!csun!polyslo!bwarkent@decvax Thu Apr 14 17:26:53 1988 In article <2130@elrond.CalComp.COM> you write: > >Upper management needs convincing that software engineers (sic) are >more productive using a multiple window windowing system over a vt100 type >(cheap) system. If you need some rederic, you might look into "The Star User Interface: An Overview." For example: "When everything being delt with in a computer system is visible, the display screen relieves the load on the short-term memory by action as a sort of `visual cache'. . . A subtle thing happens when everything is visible: the display becomes reality." Kind of a neat way of saying you could devote your high dollar brain power to solving the important problems instead of stuffing your brain with added syntax, commands, etc... brian warkentine Cal Poly, San Luis Obispo, CA ...!ihnp4!csun--\ ...!{csustan,csun,sdsu}-->--!polyslo!bwarkent ...!ucbvax!voder--/ From harvard!lilac.berkeley.edu!widow.Berkeley.EDU!adamj Fri Apr 15 10:49:19 1988 I am a convert to windows. Whenever I used to do serious programming on a character terminal, I'd always have a huge print out on each side of the terminal to look at the program and some documentation. Now my standard operating procedure is to use windows for this. The code is always up to date, and there's little effort involved in getting hold of the latest version of the programming. When I write code these days, I write it pretty fast, and can get it to compile and lint much more efficiently than before. (Having a window with the errors seperate from the editor window is a tremendous win, as is running a program and being able to edit it.) BTW, as far as cost goes, I'd like to point out that Calcomp makes a board that can drive four 1280x1024 from a microvax. Also, you can buy bitmap graphics terminals very cheaply and port X to them easily (e.g., an HDS runs ~$700, I think.) --Adam J. Richter adamj@widow.berkeley.edu Received: from savax.Sanders.COM (savax-gw.ARPA) After you find the proper research reports, you might consider asking your managers to do two gedanken experiments: 1. Take a piece of cardboard, and cut a rectangular hole in it that will allow you to read your local newspaper 24 lines at a time. Give the piece of cardboard to your managers, and ask them to spend a few days reading the newspaper with it. 2. Ask your managers to clear off their desks. Tell them they can have one piece of paper on it at a time. If they need to open a phone book, or their rolodex, they have to first put whatever else is on their desk into a drawer. I expect your managers will protest that you're being ridiculous. When they do, point out to them that you and your VT100 have worked that way for years. From ucbvax!Xerox.COM!Gary_Steinbach.osbunorth@decvax Sat Apr 16 03:44:58 1988 Julie, Hard proof is probably hard to come by, although I hope you get some quantitative responses along those lines. But when you are putting together your final pitch, don't forget that an important component of productivity is effectiveness: ending up with a higher quality product in the end. Good software tools support this in a number of ways beyond just increasing the "number of lines of code per day." These include better documentation of program goals and design, better structured (and therefore less bug-prone code), more complete and structured testing, and better management control of projects. Moreover, some systems now becoming available tie all these aspects together, hypercard style, such that if the future maintainer of a component want to know why something was implemented the way it was, she can access direct links to the appropriate design, spec, or goals document. One thing you CAN point out, even in the absence of hard, quantitative data on productivity gains, is that investments in hardware and software are relatively static; that is, they are one-time additions to budgets. The payback from that investment is multiplicative, by every work-hour spent from then on. Gary From harvard!sun.soe.clarkson.edu!nelson Sun Apr 17 00:48:05 1988 More information is available to the user at any one time. 24x80 is grossly inadequate. (he says while typing this on a 24x80 screen :-) -- char *reply-to-russ(int network) { if(network == BITNET) return "NELSON@CLUTX"; else return "nelson@clutx.clarkson.edu"; } From harvard!philabs.philips.com!mergvax!rkxyv Mon Apr 18 00:46:39 1988 Ms. Julie A. Melbin: Could you please send me any information you get that could be used to convince management that we do in fact need windowing systems ? It seems that since most of management (here at least) have never seen or used a windowing environment that the attitude is, "Well vt100's have been working find, they should continue to work fine." Thanks in advance, -Rob Kedoin From decvax!godzilla.ele.toronto.edu!ucbvax!@ai.toronto.edu:moraes Mon Apr 18 21:57:28 1988 Here's some of the points we have noticed: 1. Programming: A long window of 60-80 lines is FAR better for debugging than the old small window. Even better, a long window with a scrollbar, a la xmore... 2. Text: The previewers for troff/Tex make life *So* much easier, considering the amount of time spent doing documents around here. Turnaround time for document production has dropped considerably, since you can write something, see what it looks like, revise it, repeatedly without printouts. For a non-programming user, this is probably the single most important point. 3. You can have a separate window open for a spelling checker when writing, (ispell with asimple shell script interface) which allows you to check words as you type. 4. You can have a separate window open for man pages as you program. 5. For sysadmin work, multiple windows are wonderful - you can login to different machines, move around on the network, etc. without having to keep going ^Z, ~^Z, or suchlike. 6. Plotting utilities, (xpic for drawing pictures, plotspice for viewing graphical data, the tek window, gnuplot, are all very handy - one picture is indeed worth a thousand words). 7. The window systems offer a nice device independent graphics interface, so people have stopped writing stuff for teks, or bitgraphs, or **put your favourite graphics terminal here**. Hope these helped, Mark. PS: Please summarize to the net. Thanks. From harvard!labrea.Stanford.EDU!edsel!yduJ Thu Apr 21 04:51:24 1988 Here's some random insight about using a terminal vs. using a window system: Once I found myself in the possession of two ann arbor ambassador terminals on my desk. I hooked them both up, and found that I got a lot of the benefits of having a multiwindowing workstation. Sometimes I wished for three terminals, but don't we all wish for a little more screen real estate sometimes? In specifics: I was working on a display application, which meant that I couldn't run debugging sessions in my editor (we use gnuemacs), so with one terminal I was wasting a lot of time finding a bug, and then entering the editor to look at the the code, meanwhile forgetting the exact error message/contents of stack/procedure parameter values/whatever, and having to exit the editor to find these out and then forgetting exactly how the code went. You know the idea. But with two terminals, there was my code on one and my bug on the other and I was very happy. Having multiwindowing is better (I now have a Sun on my desk), because then I can leave a window reading the news which I look at whenever I'm waiting for a compile or whatever, and I don't have to wait for news startup. But that's really a frill. It's the two screens (sometimes three are necessary, depending on your application) that are important. And naturally if you're working on anything graphics related you need a bitmap display, and multiwindows are essential for this, so that you can look at the randomized pixels your bug has produced and at the code simultaneously. -- yduJ (Judy Anderson) edsel!yduJ@labrea.stanford.edu ...!sun!edsel!yduJ ('yduJ' rhymes with 'fudge') (415)329-8400 From ucbvax!purdue.edu!jtk@decvax Mon Apr 25 17:50:59 1988 I, too, frequently try to convince non-window users that windows are better. Would you forward me a copy of the replies you received? Tim Korb Purdue CS Department From harvard!RELAY.CS.NET!watsup.waterloo.edu!lindsay Wed Apr 27 04:38:28 1988 Second attempt to send this: In article <2137@elrond.CalComp.COM> you write: > > >A bit ago I posted a request for data to try to convince those with the >money why multi window systems are a more productive tool for engineers >than simple (old) ascii terminals. I want to thank those that have responded, >and am still looking for whatever had data/insight you want to share. Someday >we will win this battle! Would be so kind as to send me what data you have? I am trying to justify spending the money to buy some cheap X systems and would appreciate having something more than my own gut feeling to go on. You might even want to post a summary, I'm sure several people would be interested. Thanks, Lindsay Lindsay Patten Pattern Analysis & Machine Intelligence Lab lindsay@watsup Department of Systems Design Engineering lindsay@watsup.waterloo.edu University of Waterloo {uunet|utai|decvax|ihnp4}!watmath!watsup!lindsay Sorry no replys of past statistical studies, I'd sure like one. jam -- jam@elrond.CalComp.COM ...{decvax|harvard}!elrond!jam Calcomp Inc, (A Lockheed Company) Display Products Division, 65 River Road, Hudson, NH 03051-0908, Mail Stop PTP2-2D01. (603) 885 8139
jam@elrond.CalComp.COM (Julie A. Melbin) (05/12/88)
The summary I received - wish I had a study with real data too, (thanks all): Simple scenario: three tiled windows, big one running screen editor like emacs, you edit and then write the file (but don't quit emacs); second is command shell to run compiler which puts errors in file; third is where you more the error file. Never need to print listings and edit,compile, then debug is very fast. Add fourth window to view include files to get the calls correct. However, if this is not obvious to your managment, you should find another job ASAP, since the productivity and probably the products are definitely suffering from brain damaged management. A b/w Sun 3 or microVax per programmer is the minimum requirement. Anything less is a sweat shop or technically obsolete. Randall Neff@sierra.stanford.edu From harvard!SEI.CMU.EDU!sr Thu Apr 14 14:47:23 1988 Gentlemen - The acknowledged expert on empirical studies of programmers is Prof. Ben Shneiderman at the Univ of Maryland (e-mail ben@mimsy.umd.edu). Why not ask him? You can tell him that i sent you. - Stan From ucbvax!ucsd!sdsu!csun!polyslo!bwarkent@decvax Thu Apr 14 17:26:53 1988 In article <2130@elrond.CalComp.COM> you write: > >Upper management needs convincing that software engineers (sic) are >more productive using a multiple window windowing system over a vt100 type >(cheap) system. If you need some rederic, you might look into "The Star User Interface: An Overview." For example: "When everything being delt with in a computer system is visible, the display screen relieves the load on the short-term memory by action as a sort of `visual cache'. . . A subtle thing happens when everything is visible: the display becomes reality." Kind of a neat way of saying you could devote your high dollar brain power to solving the important problems instead of stuffing your brain with added syntax, commands, etc... brian warkentine Cal Poly, San Luis Obispo, CA ...!ihnp4!csun--\ ...!{csustan,csun,sdsu}-->--!polyslo!bwarkent ...!ucbvax!voder--/ From harvard!lilac.berkeley.edu!widow.Berkeley.EDU!adamj Fri Apr 15 10:49:19 1988 I am a convert to windows. Whenever I used to do serious programming on a character terminal, I'd always have a huge print out on each side of the terminal to look at the program and some documentation. Now my standard operating procedure is to use windows for this. The code is always up to date, and there's little effort involved in getting hold of the latest version of the programming. When I write code these days, I write it pretty fast, and can get it to compile and lint much more efficiently than before. (Having a window with the errors seperate from the editor window is a tremendous win, as is running a program and being able to edit it.) BTW, as far as cost goes, I'd like to point out that Calcomp makes a board that can drive four 1280x1024 from a microvax. Also, you can buy bitmap graphics terminals very cheaply and port X to them easily (e.g., an HDS runs ~$700, I think.) --Adam J. Richter adamj@widow.berkeley.edu Received: from savax.Sanders.COM (savax-gw.ARPA) After you find the proper research reports, you might consider asking your managers to do two gedanken experiments: 1. Take a piece of cardboard, and cut a rectangular hole in it that will allow you to read your local newspaper 24 lines at a time. Give the piece of cardboard to your managers, and ask them to spend a few days reading the newspaper with it. 2. Ask your managers to clear off their desks. Tell them they can have one piece of paper on it at a time. If they need to open a phone book, or their rolodex, they have to first put whatever else is on their desk into a drawer. I expect your managers will protest that you're being ridiculous. When they do, point out to them that you and your VT100 have worked that way for years. From ucbvax!Xerox.COM!Gary_Steinbach.osbunorth@decvax Sat Apr 16 03:44:58 1988 Julie, Hard proof is probably hard to come by, although I hope you get some quantitative responses along those lines. But when you are putting together your final pitch, don't forget that an important component of productivity is effectiveness: ending up with a higher quality product in the end. Good software tools support this in a number of ways beyond just increasing the "number of lines of code per day." These include better documentation of program goals and design, better structured (and therefore less bug-prone code), more complete and structured testing, and better management control of projects. Moreover, some systems now becoming available tie all these aspects together, hypercard style, such that if the future maintainer of a component want to know why something was implemented the way it was, she can access direct links to the appropriate design, spec, or goals document. One thing you CAN point out, even in the absence of hard, quantitative data on productivity gains, is that investments in hardware and software are relatively static; that is, they are one-time additions to budgets. The payback from that investment is multiplicative, by every work-hour spent from then on. Gary From harvard!sun.soe.clarkson.edu!nelson Sun Apr 17 00:48:05 1988 More information is available to the user at any one time. 24x80 is grossly inadequate. (he says while typing this on a 24x80 screen :-) -- char *reply-to-russ(int network) { if(network == BITNET) return "NELSON@CLUTX"; else return "nelson@clutx.clarkson.edu"; } From harvard!philabs.philips.com!mergvax!rkxyv Mon Apr 18 00:46:39 1988 Ms. Julie A. Melbin: Could you please send me any information you get that could be used to convince management that we do in fact need windowing systems ? It seems that since most of management (here at least) have never seen or used a windowing environment that the attitude is, "Well vt100's have been working find, they should continue to work fine." Thanks in advance, -Rob Kedoin From decvax!godzilla.ele.toronto.edu!ucbvax!@ai.toronto.edu:moraes Mon Apr 18 21:57:28 1988 Here's some of the points we have noticed: 1. Programming: A long window of 60-80 lines is FAR better for debugging than the old small window. Even better, a long window with a scrollbar, a la xmore... 2. Text: The previewers for troff/Tex make life *So* much easier, considering the amount of time spent doing documents around here. Turnaround time for document production has dropped considerably, since you can write something, see what it looks like, revise it, repeatedly without printouts. For a non-programming user, this is probably the single most important point. 3. You can have a separate window open for a spelling checker when writing, (ispell with asimple shell script interface) which allows you to check words as you type. 4. You can have a separate window open for man pages as you program. 5. For sysadmin work, multiple windows are wonderful - you can login to different machines, move around on the network, etc. without having to keep going ^Z, ~^Z, or suchlike. 6. Plotting utilities, (xpic for drawing pictures, plotspice for viewing graphical data, the tek window, gnuplot, are all very handy - one picture is indeed worth a thousand words). 7. The window systems offer a nice device independent graphics interface, so people have stopped writing stuff for teks, or bitgraphs, or **put your favourite graphics terminal here**. Hope these helped, Mark. PS: Please summarize to the net. Thanks. From harvard!labrea.Stanford.EDU!edsel!yduJ Thu Apr 21 04:51:24 1988 Here's some random insight about using a terminal vs. using a window system: Once I found myself in the possession of two ann arbor ambassador terminals on my desk. I hooked them both up, and found that I got a lot of the benefits of having a multiwindowing workstation. Sometimes I wished for three terminals, but don't we all wish for a little more screen real estate sometimes? In specifics: I was working on a display application, which meant that I couldn't run debugging sessions in my editor (we use gnuemacs), so with one terminal I was wasting a lot of time finding a bug, and then entering the editor to look at the the code, meanwhile forgetting the exact error message/contents of stack/procedure parameter values/whatever, and having to exit the editor to find these out and then forgetting exactly how the code went. You know the idea. But with two terminals, there was my code on one and my bug on the other and I was very happy. Having multiwindowing is better (I now have a Sun on my desk), because then I can leave a window reading the news which I look at whenever I'm waiting for a compile or whatever, and I don't have to wait for news startup. But that's really a frill. It's the two screens (sometimes three are necessary, depending on your application) that are important. And naturally if you're working on anything graphics related you need a bitmap display, and multiwindows are essential for this, so that you can look at the randomized pixels your bug has produced and at the code simultaneously. -- yduJ (Judy Anderson) edsel!yduJ@labrea.stanford.edu ...!sun!edsel!yduJ ('yduJ' rhymes with 'fudge') (415)329-8400 From ucbvax!purdue.edu!jtk@decvax Mon Apr 25 17:50:59 1988 I, too, frequently try to convince non-window users that windows are better. Would you forward me a copy of the replies you received? Tim Korb Purdue CS Department From harvard!RELAY.CS.NET!watsup.waterloo.edu!lindsay Wed Apr 27 04:38:28 1988 Second attempt to send this: In article <2137@elrond.CalComp.COM> you write: > > >A bit ago I posted a request for data to try to convince those with the >money why multi window systems are a more productive tool for engineers >than simple (old) ascii terminals. I want to thank those that have responded, >and am still looking for whatever had data/insight you want to share. Someday >we will win this battle! Would be so kind as to send me what data you have? I am trying to justify spending the money to buy some cheap X systems and would appreciate having something more than my own gut feeling to go on. You might even want to post a summary, I'm sure several people would be interested. Thanks, Lindsay Lindsay Patten Pattern Analysis & Machine Intelligence Lab lindsay@watsup Department of Systems Design Engineering lindsay@watsup.waterloo.edu University of Waterloo {uunet|utai|decvax|ihnp4}!watmath!watsup!lindsay -- jam@elrond.CalComp.COM ...{decvax|harvard}!elrond!jam Calcomp Inc, (A Lockheed Company) Display Products Division, 65 River Road, Hudson, NH 03051-0908, Mail Stop PTP2-2D01. (603) 885 8139