command line vs gui
in [General Linux]
|
Prev: micro solution backpack cd-writer hell
Next: "...error while loading shared libraries: libg2c.so.0"
From: blmblm on 6 Dec 2005 10:33 In article <9jeg63-kas.ln1(a)news.it.uc3m.es>, Peter T. Breuer <ptb(a)oboe.it.uc3m.es> wrote: >In comp.os.linux.misc blmblm(a)myrealbox.com wrote: >> I don't understand how one achieves "or E" at all in your interface, >> except in a rather narrow sense (e.g., name is "*.bak" or "*.tmp"). > >I was imagining that checking many checkboxes give you an OR. I'm confused .... And no wonder. Earlier you wrote: >> As I understand it, your GUI represents A & B & C & D with four >> panels, >Panels? checkboxes. One panel for all. So, does checking multiple boxes mean an AND of the conditions, or an OR? I don't think you can have it both ways! unless .... >- I >suppose that there could be an ubercheckbox saying "refine search" or >"broaden search" in order to give you the option of going either way. ..... in which case you still can't represent search criteria that combine AND and OR without -- as far as I can imagine -- more than one panel. > >If one wanted to, I suppose there could even be a "click to expand option" >place near each checkbox. > >Anyway, my only point was "this is something WELL suited to a gui, because >one constantly needs reminding of the available options and earlier >choices and one often wants to modify the preceding attempt, and the >choices of what to do are finite in number, repeated some finite number >of times" (paraphrasing). > If you want to claim that the commonest uses of "find" are well suited to a GUI, I would agree -- there are many options, and only people who use the command often will likely be able to remember them without help. You haven't yet, as far as I can tell, come up with a convincing argument for a GUI that is as expressive as the CLI, if "expressive" means "allowing the same things to be expressed, without syntax most people will find significantly more difficult than the CLI syntax." The point I'm trying to make with the last part of the condition is that anything that requires the equivalent of transforming nested expressions into conjunctive, or disjunctive, normal form is going to be difficult for many people to use. -- | B. L. Massingill | ObDisclaimer: I don't speak for my employers; they return the favor.
From: blmblm on 6 Dec 2005 10:42 In article <0teg63-kas.ln1(a)news.it.uc3m.es>, Peter T. Breuer <ptb(a)oboe.it.uc3m.es> wrote: >In comp.os.linux.misc blmblm(a)myrealbox.com wrote: [ snip ] >>>>>the atomic propositions (i.e. one window, with boxes to tick in it), >>>>>and a check box asking "finished?". > >> What does the "finished?" check box do? > >I suppose it's the opposite of "more?"! I haven't anything very >concrete in mind! Yes, I'm getting that impression .... :-)? "More"? I thought you said only one panel, with many checkboxes. [ snip ] >> How do I enter something that gives the functionality of >> "(A and B) or C"? > >Chose A, C checkboxes. Click "again". Chose B, C checkboxes. Click >"finish" ("search", whatever). Doesn't this require me to know that "(A and B) or C" can be expressed "(A or C) and (B or C)"? not necessarily in that notation, but the idea? [ snip ] >>>that covers cnf, but still). They simply click and try a search and if >>>they don't like the result they click again to modify the search - no >>>thinking (about cnf or otherwise) is required any more than you need to >>>know SQL when putting items into a google search (yes, it understands >>>AND and stuff). > >> But I claim that there are searches that cannot be done with Google, > >Sure there are - it's one of my major annoyances that you can't search >for a date range, r sort the output by date instead of their idea of >relevance. But haven't you been arguing all along that your proposed GUI doesn't restrict the power of "find"? > >> even through the "advanced" page. How do I ask for all pages that meet >> the following?: > >> ("contains foo and contains bar") or ("contains qwerty") > >To do an OR, you have to do two searches, and concat the resluts. > Exactly. No way to do it with a single search. So Google's interface understands only a subset of Boolean expressions. (I'm not saying that was the wrong choice for them to make. But if I didn't know better, I would interpret your comment that it "understands AND and stuff" to mean that it understands Boolean expressions, which -- no, only some of them.) > >> Well, maybe there's some way to tell Google "hey, I understand Boolean >> expressions, here's what I really want ...." > >There are secondary websites that do that for you. They compose an >answer for you by querying google underneath. Search for "google2" >and such. > Useful to know for next time I want that .... -- | B. L. Massingill | ObDisclaimer: I don't speak for my employers; they return the favor.
From: Peter T. Breuer on 6 Dec 2005 10:45 In comp.os.linux.misc blmblm(a)myrealbox.com wrote: > In article <9jeg63-kas.ln1(a)news.it.uc3m.es>, > Peter T. Breuer <ptb(a)oboe.it.uc3m.es> wrote: >>In comp.os.linux.misc blmblm(a)myrealbox.com wrote: >>> I don't understand how one achieves "or E" at all in your interface, >>> except in a rather narrow sense (e.g., name is "*.bak" or "*.tmp"). >> >>I was imagining that checking many checkboxes give you an OR. > I'm confused .... And no wonder. Earlier you wrote: No need to be. >>> As I understand it, your GUI represents A & B & C & D with four >>> panels, >>Panels? checkboxes. One panel for all. Correct. > So, does checking multiple boxes mean an AND of the conditions, or > an OR? I don't think you can have it both ways! unless .... An OR. Why do you think otherwise? I have never said anything to contradict that since that is the way I am imagining it! That's why I used CJF .. one has to pipe several filters together (that's an AND) so each filter must be an OR. >>suppose that there could be an ubercheckbox saying "refine search" or >>"broaden search" in order to give you the option of going either way. Here I am suggesting one let the mode (OR or AND) be chosen, since you didn't like it always being OR. > .... in which case you still can't represent search criteria that > combine AND and OR without -- as far as I can imagine -- more than > one panel. You just choose "again". >>Anyway, my only point was "this is something WELL suited to a gui, because >>one constantly needs reminding of the available options and earlier >>choices and one often wants to modify the preceding attempt, and the >>choices of what to do are finite in number, repeated some finite number >>of times" (paraphrasing). > If you want to claim that the commonest uses of "find" are well suited > to a GUI, I would agree -- there are many options, and only people who > use the command often will likely be able to remember them without > help. > You haven't yet, as far as I can tell, come up with a convincing > argument for a GUI that is as expressive as the CLI, if "expressive" Yes I have! > means "allowing the same things to be expressed, without syntax most > people will find significantly more difficult than the CLI syntax." There's no syntax involved - just click on checkboxes. > The point I'm trying to make with the last part of the condition is > that anything that requires the equivalent of transforming nested > expressions into conjunctive, or disjunctive, normal form is going to It doesn't. > be difficult for many people to use. I don't see the difficulty. What you say is required is not required. Your argument seems to boil down to "english is difficult for most people to speak because one has to translate from the chinese into english and most people can't do that". No - english is not difficult. One just "thinks in english", not in chinese. Peter
From: Peter T. Breuer on 6 Dec 2005 10:50 In comp.os.linux.misc blmblm(a)myrealbox.com wrote: > In article <0teg63-kas.ln1(a)news.it.uc3m.es>, > Peter T. Breuer <ptb(a)oboe.it.uc3m.es> wrote: >>In comp.os.linux.misc blmblm(a)myrealbox.com wrote: > [ snip ] >>>>>>the atomic propositions (i.e. one window, with boxes to tick in it), >>>>>>and a check box asking "finished?". >> >>> What does the "finished?" check box do? >> >>I suppose it's the opposite of "more?"! I haven't anything very >>concrete in mind! > Yes, I'm getting that impression .... :-)? > "More"? I thought you said only one panel, with many checkboxes. "More" just means you get the same panel to fill out again. >>> How do I enter something that gives the functionality of >>> "(A and B) or C"? >> >>Chose A, C checkboxes. Click "again". Chose B, C checkboxes. Click >>"finish" ("search", whatever). > Doesn't this require me to know that "(A and B) or C" can be > expressed "(A or C) and (B or C)"? not necessarily in that notation, > but the idea? No. > But haven't you been arguing all along that your proposed GUI doesn't > restrict the power of "find"? It doesn't. You know that. >>> even through the "advanced" page. How do I ask for all pages that meet >>> the following?: >> >>> ("contains foo and contains bar") or ("contains qwerty") >> >>To do an OR, you have to do two searches, and concat the resluts. > Exactly. No way to do it with a single search. So what? Using two google searches underneath is mere implementation! Leave that sort of thing up to the implementation. > So Google's interface > understands only a subset of Boolean expressions. As far as I know. > (I'm not saying that > was the wrong choice for them to make. But if I didn't know better, > I would interpret your comment that it "understands AND and stuff" to > mean that it understands Boolean expressions, which -- no, only some > of them.) It understands quite a few things. You can look at their help page to see what. It just doesn't understand some things that would be nice for me (like "date"), so one has to build that functionality on on top using a web interface. Peter
From: blmblm on 6 Dec 2005 12:38
In article <t8og63-0on.ln1(a)news.it.uc3m.es>, Peter T. Breuer <ptb(a)oboe.it.uc3m.es> wrote: >In comp.os.linux.misc blmblm(a)myrealbox.com wrote: >> In article <9jeg63-kas.ln1(a)news.it.uc3m.es>, >> Peter T. Breuer <ptb(a)oboe.it.uc3m.es> wrote: >>>In comp.os.linux.misc blmblm(a)myrealbox.com wrote: >>>> I don't understand how one achieves "or E" at all in your interface, >>>> except in a rather narrow sense (e.g., name is "*.bak" or "*.tmp"). >>> >>>I was imagining that checking many checkboxes give you an OR. > >> I'm confused .... And no wonder. Earlier you wrote: > >No need to be. > >>>> As I understand it, your GUI represents A & B & C & D with four >>>> panels, > >>>Panels? checkboxes. One panel for all. > >Correct. > >> So, does checking multiple boxes mean an AND of the conditions, or >> an OR? I don't think you can have it both ways! unless .... > >An OR. Why do you think otherwise? I have never said anything to >contradict that since that is the way I am imagining it! That's why I >used CJF .. one has to pipe several filters together (that's an AND) so >each filter must be an OR. > Well, you have said: >>>I was imagining that checking many checkboxes give you an OR. and also >>>> As I understand it, your GUI represents A & B & C & D with four >>>> panels, > >>>Panels? checkboxes. One panel for all. To me this second comment seems to indicate that you have checkboxes representing things (filters) to be AND'd together. Do you mean to have two kinds of checkboxes, one for options (to be OR'd together) and one for filters? (Am I really being so dense, and this is all crystal-clear to everyone else?!) > >>>suppose that there could be an ubercheckbox saying "refine search" or >>>"broaden search" in order to give you the option of going either way. > >Here I am suggesting one let the mode (OR or AND) be chosen, since you >didn't like it always being OR. > >> .... in which case you still can't represent search criteria that >> combine AND and OR without -- as far as I can imagine -- more than >> one panel. > >You just choose "again". > So now we *might* be back to representing either expressions in the equivalent of CNF, or the equivalent of DNF, or possibly one or the other. > >>>Anyway, my only point was "this is something WELL suited to a gui, because >>>one constantly needs reminding of the available options and earlier >>>choices and one often wants to modify the preceding attempt, and the >>>choices of what to do are finite in number, repeated some finite number >>>of times" (paraphrasing). > >> If you want to claim that the commonest uses of "find" are well suited >> to a GUI, I would agree -- there are many options, and only people who >> use the command often will likely be able to remember them without >> help. > >> You haven't yet, as far as I can tell, come up with a convincing >> argument for a GUI that is as expressive as the CLI, if "expressive" > >Yes I have! Well, it hasn't convinced me. It's probable that I still don't understand what you have in mind, but since I don't feel like these attempts to clarify are getting anywhere, maybe we should drop it. >> means "allowing the same things to be expressed, without syntax most >> people will find significantly more difficult than the CLI syntax." > >There's no syntax involved - just click on checkboxes. Replace "syntax" with whatever word you would use to mean "express one's meaning by clicking on the checkboxes in the GUI", okay? That was what I meant. >> The point I'm trying to make with the last part of the condition is >> that anything that requires the equivalent of transforming nested >> expressions into conjunctive, or disjunctive, normal form is going to > >It doesn't. How do you express deeply nested conditions? > >> be difficult for many people to use. > >I don't see the difficulty. What you say is required is not required. >Your argument seems to boil down to "english is difficult for most >people to speak because one has to translate from the chinese into >english and most people can't do that". No - english is not difficult. >One just "thinks in english", not in chinese. > If one is a native Chinese speaker who is attempting, as an adult, to learn English, English is difficult. "So what?" Well, I would claim that if one is fluent with the CLI for find, or with Boolean expressions in general, it will be difficult to express one's meaning with the GUI .... I'm not going to say "the GUI you describe", because I'm fairly sure I don't understand your meaning. Perhaps "the GUI I think you're describing" would be more accurate. -- | B. L. Massingill | ObDisclaimer: I don't speak for my employers; they return the favor. |