rename default file extensions to match new nomenclature?

[mr-c]

As documented in https://github.com/dib-lab/khmer/blob/doc/binaryformats/doc/dev/binary-file-formats.rst we have too many names for the same things:

1. countgraph/countinghash/count-min sketch with file extension 'ct' (counting table)
2. nodegraph/HashBits/bloom filter with file extension 'pt' (presence table)

We could

a) use the name countgraph consistently and name the files '.cg' and likewise use the name nodegraph consistently and name the files '.ng'.

b) make no changes to the file extensions

c) something else

Thoughts @kdmurray91, @blahah, @luizirber, @camillescott, @ctb, @drtamermansour (and others) ?

[ctb]

No particular attachment to .pt and .ct; those were made up by @mr-c and are inconsistently applied in any case :). I still have a sentimental attachment to .kh; maybe .kg, khmer-graph? Just .graph would probably be too confusing. We will be switching to countgraph and nodegraph internally, and I expect other types of graphs to make their way into khmer in the future, but not clear on exactly what will prove useful.

[blahah]

Given that these are going to be non-standard file extensions, wouldn't the most user friendly thing be to make them descriptive? Like maybe .count_table and .presence_table?

Living as we do in a utopian world of tab-completion, is there a benefit to having them be two letters?

[kdm9]

My brainstorming:

• Consistency FTW. +1 for .ng and .cg.
• Avoid clashes: http://filext.com doesn't list any major extension clashes for either .cg, .ct, .ng or .pt. So no preference from this angle.
• I like @blahah's suggestion (sans underscores, I'm very lazy, and we don't have underscores/camel case in the names of the data structures). But I also think that most people are comfortable with file extensions being shortened to a mnemonic few characters. The average computer user would know what to do with a .doc or a .pdf.
• I like the idea of including some abbreviation of oxli in the new extensions. maybe .oxng and .oxcg? Or something of that ilk. Or, inspired by @blahah, .oxlinode and .oxlicount?

[ctb]

On Mon, Jun 15, 2015 at 05:36:39PM -0700, Kevin Murray wrote:

My brainstorming:

• Consistency FTW. +1 for .ng and .cg.
• Avoid clashes: http://filext.com doesn't list any major extension clashes for either .cg, .ct, .ng or .pt. So no preference from this angle.
• I like @blahah's suggestion (sans underscores, I'm very lazy, and we don't have underscores/camel case in the names of the data structures). But I also think that most people are comfortable with file extensions being shortened to a mnemonic few characters. The average computer user would know what to do with a .doc or a .pdf.
• I like the idea of including some abbreviation of oxli in the new extensions. maybe .oxng and .oxcg? Or something of that ilk. Or, inspired by @blahah, .oxlinode and .oxlicount?

+1 on the last suggestions.

We're not exposing oxli to the outside world for 2.0, note, so we have some
time.

[mr-c]

@ctb is this needed now that we've started to drop auto file extension adding for our file types?

[ctb]

I think it becomes an issue only for documentation.

Titus Brown, ctbrown@ucdavis.edu

On Jul 30, 2015, at 11:12 AM, Michael R. Crusoe notifications@github.com wrote:

@ctb is this needed now that we've started to drop auto file extension adding for our file types?

—
Reply to this email directly or view it on GitHub.