Re: [Docutils-develop] Re: [ docutils-Feature Requests-1456993 ] please include rst2man/etc

[<gr...@us...>]

On Sat, 1 Apr 2006, Felix Wiemann wrote:

> gr...@us... wrote:
>
>> it fails in one point: adding one new command does not harm any old
>> code, but helps for the one case where it is used even if only
>> implemented in a small portion.
>
> Nope.  As a user, if I get a command "rst2man.py", I want it to work in
> all cases.  I don't want to rely on the fact that it works and later be
> badly surprised by feature-incompleteness, crashes, and similar funny
> things.

all cases, means all cases for a manpage not all cases for reST.
man does not support images, and i still wait for one using tables.

and if there is one user (not me) using it for his manpage, it is complete 
for this user, so what ?

>> there were several times i had to fix some sandbox things because the
>> api changed, what do you expect you would need to fix if i add some
>> new writer.
>
> Well, we'd have to keep the man page writer up-to-date.  That can be
> difficult if the code isn't clean, so I don't want to commit to it as
> long as it isn't complete, reasonably bug-free, tested, and documented.

i did keep the latexwriter uptodate, so why do you imply i wouldnt do it 
with a manpage writer.

>> commiting to the main trunk does not hit any normal user, only
>> snapshot users,
>
> Snapshot users *are* normal users.  (We encourage users to use the
> snapshot on the Docutils homepage!)
>
> Also, if we allow code unsuitable for normal users in the trunk, we
> cannot follow the "release early, release often" policy anymore.  I want
> the trunk to be releasable at any time.

1. who says i code unusable for normal users ? 
2. i am for "release early, release often" therefore i would put it into
    main trunk early, not late.

the problem starts here

>> but makes it easier for developers to just run the
>> command, without the need to checkout a separate branch.
>
> Well, I have a complete checkout of the whole repository (trunk,
> branches, and tags).  And you can always use "svn switch".

does alltest.py work over branches ?
getting into

>> anyway the project is not urgent, so i could start a sandbox with
>> manpage-writer or make a branch.
>
> Unless you expect to get the writer feature-complete quickly, I'd
> recommend using the sandbox because a branch would diverge from the
> trunk over time.

correct

>> but in a sandbox it is harder to reuse the testing environment
>
> Now this is a problem.  Should we maybe move the tests into the main
> docutils directory, so that you can say "from docutils.test import
> DocutilsTestSupport"?  I'm not sure...

now you see, add to it that it is in a drifting branch

>> and one needs a setup.py,
>
> Mh, yes.  The bad thing is that we don't have a plugin mechanism yet,
> otherwise you could just drop the package in ~/.docutils/plugins/.

and that users cant use it

>>>> using a branch allows every developer to try out, but no user.
>
> We can implement daily (or hourly) snapshots for branches if the need
> arises.
(not really, if there is a user complain/feature request i usually mail 
the source and commit.)

and we need more snapshots

you have a little much effort for ... uhm nothing, the only thing that 
can happen, is that some user tries the rst2man.py and it does not work, now 
there are complains on rst2latex too should i remove it therefore ?

i understand your and davids arguments, but they are somehow hindering 
contributing. did anyone recognize that i am the only one that took a 
writer out of the sandbox. on users people keep asking for stuff that is 
in the sandbox, but out of sight.


cheers

--