Newline honor dedent

If py-newline-and-indent is called from an empty but dedented line, don't go to the outmost indent, honor dedent.

Thank you very much, this seems to solve the problem when indent is decreased manually.

I only wonder if newline should honor as well absence of any indent in an empty line, that is of course in case when the user removes all the indent on an empty line and then presses return? Currently the cursor would be moved again to the outmost indent.

Additionally, probably there is a little typo in the name py--delete-trailing-whitepace -- should it be py--delete-trailing-whitespace?

Am 17.02.2014 13:51, schrieb lukas:
> Thank you very much, this seems to solve the problem when indent is
> decreased manually.
>
> I only wonder if newline should honor as well absence of any indent in
> an empty line, that is of course in case when the user removes all the
> indent on an empty line and then presses return? Currently the cursor
> would be moved again to the outmost indent.
>

Thanks, see there is something left. Also if last command was a indent a certain pos, probably that should be honored.

> Additionally, probably there is a little typo in the name py--delete-
> trailing-whitepace -- should it be py--delete-trailing-whitespace?
>

Yes, will fix it.

Thank you very much! The dedent now works great. But just noticed one thing which would require a fix probably in this place.

Whenever there is an empty line after a block of code - let's say for the reproduction purpose: the last line in the block has 4 levels of indention - and the user introduces an indent different in the empty line following the block and presses RET, the point is moved to a new line and again moved to the outmost indent. This does not seem being disruptive and might be even desired, but it gets under a question as a consequence in this case:

[0] let's start from beginning

[1] after a line with an indent n, there is a "second" non empty line with indent m < n

[2] the "second" line can be preceded by any amount of empty lines (zero or hundred empty lines included)

[3] the user places the point at the beginning of the "second" line or on the first non whitespace character in the "second" line and presses RET

[4] again the point is moved to next line and to the outmost indent, with the whole text following the cursor (***) -- I think that in such a situation the desired behavior would be to just move the text to next line with point and first char of the text placed precisely in the same column

*** if the point has been placed on the beginning of the "second" line, and not on the first non whitespace character, that is when there was a whitespace between the point and the following first non whitespace character of the "second" line, after pressing RET, the following text not only will be moved to the outmost indent on next line, but it will be moved with all the additional whitespace in front of it, resulting in even more increased indent

Am 18.02.2014 16:34, schrieb lukas:
> Thank you very much! The dedent now works great. But just noticed one
> thing which would require a fix probably in this place.
>
> Whenever there is an empty line after a block of code - let's say for
> the reproduction purpose: the last line in the block has 4 levels of
> indention - and the user introduces an indent different in the empty
> line following the block and presses RET, the point is moved to a new
> line and again moved to the outmost indent.

That would mean a bug.
Implementation is now: if user indents an empty line deliberately, this indent shall be honored by a following py-newline-and-indent.

May you send some example code, which triggers the bug?

Linking to colored version as it might provide a better overview:
http://paste.debian.net/82756/

Hope this helps :)

Am 18.02.2014 19:08, schrieb lukas:
> Linking to colored version as it might provide a better overview:
> http://paste.debian.net/82756/
>
> Hope this helps :)
>

Hmm, the new feature will take action only, if a dedent command precedes.
If you just jump onto some spot which is indented already, it indents outmost.

Might be reasonable to cancel the limitation, will re-open.