| sberube | That could fly |
| whatspy | laughing |
| peaker | sberube: nope, programming via wysiwyg |
| hays | A corrolary to your statement would be that lower-level languages are less optimizable, e.g. faster without optimization. |
| peaker | like in www.subtextual.org |
| deitarion | More like wywiwyg...wdadt (What you want is what you get... without doing a damn thing) |
| peaker | hays: they may be faster without optimization and they are (e.g C) but the higher ones are much faster because an optimizer can do a much better job than a human _feasibly_ can on a large project deitarion: You still have to create the semantic model, you just edit semantics, not syntax via specialized code editor widgets and you can have multiple views of the same semantic model. Viewing/not-viewing the static types is a key/checkbox away |
| whatspy | hmmm. sounds like a vast abstraction |
| deitarion | I'd have to think about that. Depends on how well it lends itself to exploration of the on-disk representation of the program. |
| hays | Peaker: I'm not sure I would call Fortran higher level than C. But it is more optimizable, mainly due to the rules imposed on artithmetic evaluation. |
| peaker | whatspy: It "feels" less abstract, because it shows all of the code running on examples, and all intermediate results (see the demo on www.subtextual.org) hays: in C you can do too many things that there's no way the compiler can figure out, so its optimizability is very limited |
| sberube | Peaker: This would actully only be useful in certain special circumstances. |
| hays | Fortran has less guarantees of evaluation order and can therefore optimize math better |
| peaker | I don't really know Fortran to say much about it sberube: what would? |
| sberube | Peaker: subtext |
| peaker | hays: "less restriction on how to implement it" does make a language higher it is possible Fortran is higher than C, even if it sucks :-) sberube: why do you think so? |
| sberube | Peaker: Fortran and C are about the same level |
| CruJones | is there a way of not showing the raw_input of a host in the terminal (password) |
| sberube | Peaker: Most scientific programming and embedded programming depend a lot on knowing machine logic. Abstraction layers are built on top of that. No matter how you go around it for a high level language you will lose flexibility if you try to use subtext Peaker: So in short it would be very useful for simple programs and very high level scripts. But it could never be a real programming language to replace what we know |
| hays | ok I need to get back to work |
| peaker | sberube: "Knowing machine logic" -- what machine logic? |
| sberube | Peaker: machine code, operations, memory addressing |
| hays | Well Fortran manages to ignore most of that |
| peaker | sberube: You could make "sub-languages" in a subtext-world, which are different semantic models, that you edit with different widgets, and write a different compiler for, and translate to native code |
| whatspy | have you read Joel on software's "leaky abstractions" post ? |
| sberube | sberube: Well you could hack around it. But the fact is the current implementations are really the simplest and most robust... it wouldnt make sense to change them for something more complex |
| peaker | sberube: You still gain the advantages of semantic editing, a lot of existing functional stuff can be easily compiled to such a semantic model directly, so these semantic objects will be able to use the other semantic models via something like FFI whatspy: Yes, I think he's right about abstractions simplifying code but not the required understanding, but I think he's wrong about calling it "leaks". TCP _defines_ that in the case of under-layer failure it will propagate those failures |
| sberube | Peaker: You are thinking way too high level. High level is useful and nice for some things but the best programming languages always have a loophole to go back to lower level for special cases. In this case this subtext would be too different from the lower level to make it actually useful. Hence adding complexity |
| CruJones | how do i get the username of the person connected to my socket like how ssh does it |
| peaker | sberube: Its not about high or low level |