4 ms·

Example usage from the readme below. There is no explanation whatsoever provided as to what is going on with all the different formats. Build up test data

by natch 8y ago

Example usage from the readme below. There is no explanation whatsoever provided as to what is going on with all the different formats. Build up test data for low level code. $ bo -n oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011 0x3f, 0xc0, 0x00, 0x00, 0x3f, 0xa0, 0x00, 0x00, 0x03, 0xe8, 0x07, 0xd0, 0x0b, 0xb8, 0xff, 0xfe, 0x7a, 0x8b What? What kind of value starts with a lower case ‘o’ or a ‘P’? Or are those commands? If so, why don’t they start with hyphens? Are they file names? If so, why aren’t they preceded by -i or -o to match the docs? Is some of it output? If I used this, how would I explain my work to my coworkers? A tool should solve problems, not give me a new ones. My approach is usually to make a tool friendly enough that a newbie can glance at a command typed in by anyone and immediately see what it is supposed to be doing, without reading any help file. But in this case even the help file doesn’t help.

kstenerud 8y ago
The HN crowd are my UX testers. Whatever people don't understand is easily fixed with a git push!
- mmastrac 8y ago
  I'll attempt to possibly convert the grandparent comment into something more constructive. I'd note that your tool is probably the appropriate complexity level compared to utilities of the same sort (ie: sed, awk, dd, etc). But let's say we want to improve the UX of the tool for new users. As a strawman, you could offer longhand forms of the mode switching flags for them. For example: bo -n oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011 Could be written long-form as: bo -n --output=type=hex,width=1,endian=l,width=2 --preset=c --input=type=float,width=4,endian=b 1.5 1.25 --input=type=int,width=2,endian=b 1000 .... Each of your commands could have an advanced (oh1l2), medium (-o h1l2 or --output h1l2) or beginner (--output=type=hex,width=1,endian=l,width=2) syntax. You could even print the shorthand version to stderr whenever one of the simpler syntaxes was used.
  - kstenerud 8y ago
    That's actually another part that fell between the cracks and I'm attempting to address... Aside from the -n, everything is parsed as a whitespace-delimited stream. They just happen to be passed as command line arguments, but they're really just stream data. So all the input, output, formatting, and data commands can be passed as one big string argument bo -n "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011" Or as multiple arguments (no point here, just that you can do it): bo -n "oh1l2 Pc if4b 1.5 1.25" "ii2b 1000 2000 3000" "ih1 ff fe 7a ib1 10001011" Or as a file from disk: echo "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011" >mycommands.txt bo -n -i mycommands.txt Or as multiple files from disk: echo "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000" >mycommands1.txt echo "3000 ih1 ff fe 7a ib1 10001011" >mycommands2.txt bo -n -i mycommands1.txt -i commands2.txt Or as stdin: echo "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011" | bo -n -i - Or as a crazy mix of all of that: echo "if4b 1.5 1.25" >float-32s.txt echo "ii2b 1000 2000 3000" >int-16s.txt echo "ih1 ff fe 7a" >hex-bytes.txt echo "ib1 10001011" >binary-bytes.txt bo -n -i float-32s.txt -i int-16s.txt -i hex-bytes.txt -i binary-bytes.txt oh1l2 Pc It's designed to fit in with the unix file and pipe paradigm, and also be composeable so that you can have saved data chunks to use in multiple cases, which by necessity makes it a bit opaque for someone approaching it for the first time, but also makes it incredibly powerful when mixed with other unix tools.
    - oliverevans96 8y ago
      You might be interested in docopt (http://docopt.org/ http://docopt.org/), which is a multi-language command line parser that removes almost all of the boiler plate. Whereas the versions of docopts for Python, etc. just require an import, the C version emits C code to do the parsing. It may be worth a look!

by natch 8y ago

kstenerud 8y ago

The HN crowd are my UX testers. Whatever people don't understand is easily fixed with a git push!

mmastrac 8y ago
I'll attempt to possibly convert the grandparent comment into something more constructive. I'd note that your tool is probably the appropriate complexity level compared to utilities of the same sort (ie: sed, awk, dd, etc). But let's say we want to improve the UX of the tool for new users. As a strawman, you could offer longhand forms of the mode switching flags for them. For example: bo -n oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011 Could be written long-form as: bo -n --output=type=hex,width=1,endian=l,width=2 --preset=c --input=type=float,width=4,endian=b 1.5 1.25 --input=type=int,width=2,endian=b 1000 .... Each of your commands could have an advanced (oh1l2), medium (-o h1l2 or --output h1l2) or beginner (--output=type=hex,width=1,endian=l,width=2) syntax. You could even print the shorthand version to stderr whenever one of the simpler syntaxes was used.
- kstenerud 8y ago
  That's actually another part that fell between the cracks and I'm attempting to address... Aside from the -n, everything is parsed as a whitespace-delimited stream. They just happen to be passed as command line arguments, but they're really just stream data. So all the input, output, formatting, and data commands can be passed as one big string argument bo -n "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011" Or as multiple arguments (no point here, just that you can do it): bo -n "oh1l2 Pc if4b 1.5 1.25" "ii2b 1000 2000 3000" "ih1 ff fe 7a ib1 10001011" Or as a file from disk: echo "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011" >mycommands.txt bo -n -i mycommands.txt Or as multiple files from disk: echo "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000" >mycommands1.txt echo "3000 ih1 ff fe 7a ib1 10001011" >mycommands2.txt bo -n -i mycommands1.txt -i commands2.txt Or as stdin: echo "oh1l2 Pc if4b 1.5 1.25 ii2b 1000 2000 3000 ih1 ff fe 7a ib1 10001011" | bo -n -i - Or as a crazy mix of all of that: echo "if4b 1.5 1.25" >float-32s.txt echo "ii2b 1000 2000 3000" >int-16s.txt echo "ih1 ff fe 7a" >hex-bytes.txt echo "ib1 10001011" >binary-bytes.txt bo -n -i float-32s.txt -i int-16s.txt -i hex-bytes.txt -i binary-bytes.txt oh1l2 Pc It's designed to fit in with the unix file and pipe paradigm, and also be composeable so that you can have saved data chunks to use in multiple cases, which by necessity makes it a bit opaque for someone approaching it for the first time, but also makes it incredibly powerful when mixed with other unix tools.
  - oliverevans96 8y ago
    You might be interested in docopt (http://docopt.org/ http://docopt.org/), which is a multi-language command line parser that removes almost all of the boiler plate. Whereas the versions of docopts for Python, etc. just require an import, the C version emits C code to do the parsing. It may be worth a look!