Talk:GNU Units

Sorry
I moved this page from Units (computer program) to GNU Units, as I thought it was only referring to "GNU Units", however it appears it also refers to "units" from AT&T. I've also added UDUNITS, which pre-dates GNU Units (based on ftp timestamps .. so this interpretation could be off). Please feel free to move it back to Units (computer program) if you wish. +mwtoews 01:04, 3 November 2006 (UTC)
 * I've moved this to a more generic Units (Unix), which fits much more like the others in List of Unix programs. +mwtoews 09:22, 5 November 2006 (UTC)

Edits of 23–30 April 2011
I put GNU units in a separate section to make it more clear that the material, especially the “extensively commented” data file, applied only to GNU units (the file with the original version was uncommented).

The example I added was done with version 1.88. Strictly, it probably should be shown as a separate invocation that begins with the announcement



but I think it’s less confusing as I have it. If we think it necessary to be strictly correct, we could rerun the previous examples with version 1.88. JeffConrad (talk) 10:09, 23 April 2011 (UTC)

I cannot find any definition of furlong that makes it 0.12500025 mile, so I re-ran all examples using GNU units version 1.88. I changed the first command-line example to show a typical prompt with Windows cmd.exe to demonstrate that the program really is cross platform. The second command-line example was re-run with a C shell on Windows XP.


 * Documents on the NIST Web site suggest that the furlong is based on the US survey foot rather than the international foot, so the two examples I changed well may be correct. I’ve raised the issue on Talk:Furlong. The GNU units program may also need revision; the author is out of town, so it will be several weeks before this can be addressed. In any event, I think this clearly demonstrates the need to give the specifics for the program used to produce the examples. JeffConrad (talk) 09:37, 28 April 2011 (UTC)


 * The Weights and Measures Act 1985 gives the furlong in terms of the international yard, equal to 0.9144 metre, suggesting that the official definition in the UK is different from that in the US. GNU units defined the furlong in terms of the U.S. survey yard prior to version 1.84d, but then changed the definition to be in terms of the international foot. For now, I’ve left the example as it stands to reflect the current version of GNU units. I’ve added a note mentioning the change in definitions so users of the versions prior to 1.84d won’t be wondering what’s happening. JeffConrad (talk) 11:47, 30 April 2011 (UTC)

Because of some of the above differences, I included the specific program description in the heading; if it adds too much clutter, remove it and rely on the comments. JeffConrad (talk) 02:30, 25 April 2011 (UTC)

“Complex units expressions” example
I put the “complex units expressions” example in a separate section to give it some context; I think it’s an application that’s not always obvious. It may be approaching a “how to”; revert if it goes too far.

The original example (which I managed to mangle in transposing from a pressure-drop solution to a flow solution) was also a form of the Darcy–Weisbach equation, and because of its added complexity, was arguably better illustrative of the benefit of using units to obtain a conversion constant for an equation not using consistent (e.g., SI) units. But the corresponding form of the equation,



\bar P \Delta P = \frac {8} {\pi^2} \frac {m} {kT_\mathrm{s}} \frac {T} {T_\mathrm{s}} fL                 \frac {Q^2_\mathrm{s}} {d^5} \,, $$

is sufficiently uncommon that I don’t think it could reasonably be included without a source (which I don’t have) or a derivation (which isn’t appropriate for this article). JeffConrad (talk) 07:59, 5 May 2011 (UTC)

Edit of 29 March 2014—update for version 2.10
I’ve updated the article to reflect units version 2.10, and made some general editorial changes to the rest of the article.
 * Added mention of the source in the original style from the Heirloom Project. The features are limited compared to the current version, but in its time, it was quite a useful program.
 * Added brief summaries of changes in versions 2.10 and 2.02. I don’t think it’s worth going much further because the changes are covered in great detail in notes included with the source distribution.
 * Attempted to clean up the description of changes in version 2.00; in some cases, I didn’t completely understand the explanation (and my German is so pathetic that I didn’t attempt to translate the cited article). I hope I’ve captured the intent of the previous comments.
 * Changed the section order under History so that “Other” follows the two specific implementations.
 * Expanded the description of the differences between GNU units and the original Unix version.
 * Corrected an erroneous statement I had made that the original Unix version allowed only one solidus; although it largely worked to that effect, it wasn’t strictly correct. The full explanation is a bit involved, and including it didn’t seem indicated; it’s described in the program documentation under “Operators”.
 * Expanded the examples to illustrate some of the new features; all examples were run using version 2.10 in the environments depicted. I didn’t include any examples using the Heirloom version because I didn’t think it would add much, but it’s easy enough to do if someone thinks it would be useful.
 * Removed the comment about the new definitions of furlong and related units; I had added this comment several years ago when I changed the examples to use the then-current version; it has now been quite some time since the definitions were revised, and I think the comment now served more to clutter than enlighten. The practical significance is usually insignificant; the main effect is to preserve the familiar integral relationships among the foot, rod, chain, furlong, and mile.
 * Removed the syntax highlighting from the examples. The highlighting was somewhat misleading for examples run from a command prompt or interactively in the program. Moreover, the indicated languages were incorrect; I’ll concede that “bash” would probably suffice for a Korn shell script, but using a “dos” tag for the cmd.exe example gives very strange results. And again, the highlighting seemed to suggest, at least at first glance, that the entries were actually scripts.
 * Added the GNU units webpage to External Links because it’s arguably the most relevant to this article. It’s already included in the Infobox in the lead section, but that link is easy to overlook.
 * Revised the title for the link to the GnuWin32 so as not to imply that it’s the only Windows port.
 * Updated the broken link to the NetCDF Climate and Forecast Metadata Conventions.

The article relies heavily on primary sources (the GNU documentation and release notes), but I honestly don’t expect a review by CNET or the New York Times in the near future. Full disclosure: I’ve had periodic involvement in the project for many years, and especially in the last several releases.

JeffConrad (talk) 23:11, 29 March 2014 (UTC)