insertPackage could accept "fields" argument passed to write_PACKAGES

[jangorecki]

So user can have better control over what metadata are being part on the PACKAGES index file.

[eddelbuettel]

The idea of opening a issue ticket is to give me a moderate and reasonable amount of time to actually look at the ticket, read it and comment.

Immediately following up with PR is useless.

Please motivate why all users of drat need a new field.

[eddelbuettel]

A priori I am not in favour of doing anything which R itself does not do. One key argument for drat is that everything is as plain vanilla as possible, and that the exact same tools are used.

[jangorecki]

but this is what R do tools::write_PACKAGES. Why not use metadata from DESCRIPTION file to build PACKAGES index? I would say this is a feature of base R / tools vs drat at the moment.

[eddelbuettel]

oh, I see -- you got a point there. makes it way more elegant and I am starting to warm up to it.

patch then lets you clobber your PACKAGE with whatever you want to put, everybody else gets what they always had and we have a path forward.

I think we can do that. lemme give it a day.

[jangorecki]

should be doable with ... instead of new argument

[eddelbuettel]

or new (to us) argument with default value that is set, provided we don't mess with the order

[jangorecki]

Fields expected by default won't be reordered, new lines from provided fields are after the defaults. What do you think about ... in 66e407a?

[eddelbuettel]

looks good. Is ...=... legit? Don't think I have seen that...

[jangorecki]

While ...=... is legit, it shouldn't be here, as write_PACKAGES doesn't have ... arg. Updated in latest commit.

[eddelbuettel]

Well caught, thank you.