Hello Yakov, As far as I know, there is no convention that would force either text/comma-separated-values or text/cvs. My suggestion is to pick one, and have implementations converge on it. I'd personally prefer text/cvs, just because it's shorter. Regards, Martin. At 08:08 05/02/04, Yakov Shafranovich wrote:

Martin,

The reason why I put in an additional 'text/comma-separated-values' type is because there is an existing registration for "text/tabbed-separated-values" and I thought that the CSV one should correspond to the existing one. However, its too long for practical use which is why I included "text/csv" as well. In practice I have seen both being used.

In truth, I don't know enough about the historical conventions for MIME types to judge whether an existing registration for "tabbed-separated-values" should mandate "comma-separated-values". I assumed that there is a convention, but if in fact I am wrong, I will be happy to adjust the draft to target one type.

As for the encoding section, you have a valid point. I will adjust it for the -01 draft including the ABNF grammar in the appendix.

Yakov

Martin Duerst wrote:

...

Hello Yakov, Good work. Two comments: - I think registering just text/cvs should be enough. One point of the registry is to reduce the number of values used for the same thing to one. The registration can serve to converge to that.

...

- I like the realism in the "Encoding considerations" section, but shouldn't you at least mention LF also for line ends, or just leave out specific values? Regards, Martin.

At 10:58 05/02/03, Yakov Shafranovich wrote:

...

Hi all,

I noticed that the MIME registration tree has an entry for 'text/tab-separated-values' but for some reason doesn't have any registration for 'text/comma-separated-values' or 'text/csv. I have been reviewing some of the stuff out there and so far have seen the following being used:

application/csv text/csv text/comma-separated-values text/x-csv

I was wondering if there would be any value to formally register two of the CSV types being used from the text tree (text/csv and text/comma-separated-values). I put together a small draft which hasn't posted to the ID repository yet. A copy should be available here:

http://www.shaftek.org/publications/drafts/mime-csv/draft-shafranovich-mime-...

...

...

Yakov

At 06:06 05/02/06, Bruce Lilly wrote:

On the other hand, "comma-separated-values" is meaningful, whereas neither the originally-proposed "csv" nor Martin's suggested "cvs" is; and that probably explains the transcription error.

Appologies for my typo. I think "csv" is known well enough for people who use it frequently enough. It's also the customary extension. So I don't see any problems to go with "csv". Regards, Martin.