--===============3915771384207252567== Content-Type: multipart/alternative; boundary=047d7b6d878a84d34204eafd66e9 --047d7b6d878a84d34204eafd66e9 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: quoted-printable Right, that was my point. With something like iPython, it shouldn't be much more difficult to support both python 2 and python 3 than just one or the other, you just need either the user or the distro to set the right command to execute. Supporting two versions wouldn't be any more difficult than just allowing two commands to be set (which for something like iPython, which has a lot of flexibility with different profiles, might be a good thing even without python 2 support). On Tue, Nov 12, 2013 at 2:12 PM, Philipp A. wrote: > =93link against it=94? it=92s no native library, so that wouldn=92t even = be > possible. > > but no, we don=92t hard-depend on it: it=92s just possible to load a P=E2= t=E9 > plugin which depends on it, and embeds a =93IPython Qt Console=94 widget = into > Kate=92s bottom Toolview area. Like this > > > 2013/11/12 Todd > >> On Mon, Nov 11, 2013 at 5:57 PM, Sven Brauch wrote: >> >>> On Monday 11 November 2013 12:55:06 J. Pablo Mart=EDn Cobos wrote: >>> > Because the pep8, >>> > pyflakes and parse syntax checker will report errors, because the >>> syntax is >>> > different in python 2 and python 3. >>> Okay, I looked a bit more and this is indeed a valid concern. >>> >>> The problem is that some of the plugins (pep8 and the ipython console, = at >>> least) call straight into the mentioned tools and do something with >>> them, and >>> the tools (e.g. ipython console) will be the version of the language >>> kate was >>> linked against. >>> >>> For most of the tools the solution is imo easy: e.g. pep8 should just b= e >>> called as a program and the output should be parsed. It's very simple >>> ... yes, >>> not as elegant, but who cares. >>> >>> For ipython however it's less obvious ... I'm not sure how the problem >>> could >>> be solved here. The only solution I found is based on creating a widget >>> and >>> then letting a new process draw into the widget, but that sort of sucks >>> (especially I'm not sure how portable it is). That's more work and more >>> bad >>> than just keeping the current situation, so not worth it imo. >>> >>> What I said earlier in the thread was referring only to what language >>> should >>> be available for writing the plugins, assuming plugins which are just >>> doing >>> stuff but not actually _importing python tools and using them >>> in-process_. >>> For this issue, I'm not sure how to proceed. Possibly the "link against >>> both >>> libs" is indeed the best solution. >>> >> >> Why, exactly, do we need to link against iPython? Why can't there just >> be a console that launches iPython, or an html canvas that iPython write= s >> to? >> >> _______________________________________________ >> KWrite-Devel mailing list >> KWrite-Devel@kde.org >> https://mail.kde.org/mailman/listinfo/kwrite-devel >> >> > > _______________________________________________ > KWrite-Devel mailing list > KWrite-Devel@kde.org > https://mail.kde.org/mailman/listinfo/kwrite-devel > > --047d7b6d878a84d34204eafd66e9 Content-Type: text/html; charset=windows-1252 Content-Transfer-Encoding: quoted-printable
Righ= t, that was my point.=A0 With something like iPython, it shouldn't be m= uch more difficult to support both python 2 and python 3 than just one or t= he other, you just need either the user or the distro to set the right comm= and to execute.=A0 Supporting two versions wouldn't be any more difficu= lt than just allowing two commands to be set (which for something like iPyt= hon, which has a lot of flexibility with different profiles, might be a goo= d thing even without python 2 support).

On Tue, Nov 12, 2013 at 2:12 PM, Phili= pp A. <flying-sheep@web.de> wrote:

=93link again= st it=94? it=92s no native library, so that wouldn=92t even be possible.

but no, we don=92t hard-depend on i= t: it=92s just possible to load a P=E2t=E9 plugin which depends on it, and = embeds a =93IPython Qt Console=94 widget into Kate=92s bottom Toolview area= . Like this



2= 013/11/12 Todd <toddrjen@gmail.com>
On Mon, Nov 11, 2013 at 5:57 PM, Sven Brauch <svenbrauch@googlema= il.com> wrote:
On Monday 11 November 2013 1= 2:55:06 J. Pablo Mart=EDn Cobos wrote:
> Because the pep8,
> pyflakes and parse syntax checker will report errors, because the synt= ax is
> different in python 2 and python 3.
Okay, I looked a bit more and this is indeed a valid concern.

The problem is that some of the plugins (pep8 and the ipython console, at least) call straight into the mentioned tools and do something with them, a= nd
the tools (e.g. ipython console) will be the version of the language kate w= as
linked against.

For most of the tools the solution is imo easy: e.g. pep8 should just be called as a program and the output should be parsed. It's very simple .= .. yes,
not as elegant, but who cares.

For ipython however it's less obvious ... I'm not sure how the prob= lem could
be solved here. The only solution I found is based on creating a widget and=
then letting a new process draw into the widget, but that sort of sucks
(especially I'm not sure how portable it is). That's more work and = more bad
than just keeping the current situation, so not worth it imo.

What I said earlier in the thread was referring only to what language shoul= d
be available for writing the plugins, assuming plugins which are just doing=
stuff but not actually _importing python tools and using them in-process_.<= br> For this issue, I'm not sure how to proceed. Possibly the "link ag= ainst both
libs" is indeed the best solution.

Why, exactly, do we need to link against iPython?=A0 Why can't t= here just be a console that launches iPython, or an html canvas that iPytho= n writes to? =A0

_________________________________________= ______
KWrite-Devel mailing list
KWrite-Devel@kde.= org
https://mail.kde.org/mailman/listinfo/kwrite-devel



_______________________________________________
KWrite-Devel mailing list
KWrite-Devel@kde.org
https://mail.kde.org/mailman/listinfo/kwrite-devel


--047d7b6d878a84d34204eafd66e9-- --===============3915771384207252567== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ KWrite-Devel mailing list KWrite-Devel@kde.org https://mail.kde.org/mailman/listinfo/kwrite-devel --===============3915771384207252567==--