Is --parser necessary?

[ryuheechul]

Yesterday (well I was introduced to dasel yesterday!), I was confused with -p option and you can read about what mistake I made in here, #153.

Now I know -p is shorthand for --parser, thanks to the answer in the issue above 😄 .

However I think it's possible to get rid of -p all together if we use -r (--read) value to result as a default value to print what output format to print. (Which I think what most people would expect to happen)

For example,
Currently echo "who: i'm yaml" | dasel -p yaml should just work with echo "who: i'm yaml" | dasel -r yaml.
And in fact, it already does that now. (same with echo '{"who": "i am json"}' | dasel -r json)

When you need to change the output to json, you will need to echo "who: i'm yaml" | dasel -r yaml -w json anyway.
So I don't see -p is doing what other options can't in both situations above.

Thus, to the original point. Why do we need --parser?

[TomWright]

The ability to convert between data types was added in v1.7.0: https://github.com/TomWright/dasel/blob/master/CHANGELOG.md#v170---2020-11-30

Prior to this the -p flag was used to tell dasel which data type to expect when reading from stdin.

In order to keep backward compatibility the flag has to remain but as you say, you can essentially forget about it and just use -r since internally when no -w flag is provided, the -r value is used.

• -r is shorthand for --read, which is used to set a read parser.
• -w is shorthand for --write, which is used to set a write parser.
• -p is shorthand for --parser, which is effectively an alias of -r <val> -w <val>