On Sunday, November 23, 2014 11:36:02 AM UTC, Jeroen Demeyer wrote:
>
> 1) Keep preparse() in Sage as it is.
> 2) Use IPython, which calls preparse(), to keep track of the relation
> between unpreparsed and preparsed source code.
> 3) Don't try to change the raw traceback (as returned by the Pyt
On 2014-11-23 12:11, Volker Braun wrote:
Whoever puts stuff last in the line cache wins, obviously. IPython
obviously puts the current input there just before executing.
OK, for the first time in this thread I think I actually understand your
argument. Am I right that you're arguing:
1) Keep
On Sunday, November 23, 2014 10:53:44 AM UTC, Jeroen Demeyer wrote:
>
> > Also, the commandline interface would then have to override IPython's
> > use of the linecache.
> I don't see why. You can have two libraries (Sage and IPython) both
> using the Python linecache library, why is that a prob
On 2014-11-22 22:53, Volker Braun wrote:
IPython is, at its core, a library to apply transformations to
sources
No, sage.misc.preparser.preparse() applies transformations, IPython just
calls that function.
and evaluate them
A one-liner in Python, no need for IPython here.
with easy-to-read
On Saturday, November 22, 2014 6:05:47 PM UTC, Jeroen Demeyer wrote:
>
> but then you're essentially forced to tie (un)preparsing closer to
> IPython. I doubt that this is the right thing to do.
Why not? IPython is, at its core, a library to apply transformations to
sources and evaluate them wi
On 2014-11-22 13:51, Volker Braun wrote:
IPython colorizes the tracebacks (at least by default, unless you turn
it off) and adds the source line for command-line input. I think
unpreparsing could easily be added on top of that:
Of course it *could* be done (assuming upstream accepts it of course)
On Saturday, November 22, 2014 8:53:16 AM UTC, Jeroen Demeyer wrote:
>
> > 5) beautify the traceback
> I assume with "beautify" you mean the graphical layout of the traceback?
> Because IPython doesn't really change the contents of the traceback
> (like unpreparsing).
>
IPython colorizes the t
Let me explain further:
#71 is mostly about displaying unpreparsed code in tracebacks.
Currently, IPython doesn't do this. Given that we also need to support
this for code from load() and attach(), I think the following three
should be implemented in the same place (either all in Sage or all i
On 2014-11-21 18:21, Volker Braun wrote:
Its not brain surgery, but you want
1) Apply the preparser, possibly other input transformations
2) handle syntax errors from the preparser and show an appropriate error
(not: a traceback inside the preparser)
3) handle errors from the string -> ast com
Its not brain surgery, but you want
1) Apply the preparser, possibly other input transformations
2) handle syntax errors from the preparser and show an appropriate error
(not: a traceback inside the preparser)
3) handle errors from the string -> ast compilation and show an appropriate
error
On 2014-11-21 15:29, Volker Braun wrote:
You don't have to use IPython interactively, just use its machinery for
evaluating code-from-a-string instead of reinventing the wheel.
What's wrong with exec(compile())? There is not much reinvented in that.
Note that it's not just code evaluated from a
You don't have to use IPython interactively, just use its machinery for
evaluating code-from-a-string instead of reinventing the wheel. Really, why
doesn't SageNB just do the equivalent
of sage.repl.interpreter.get_test_shell().run_cell('1+1'), instead there is
multiple temporary files with som
On 2014-11-21 14:00, Volker Braun wrote:
IMHO this would better be solved upstream.
I don't want to make this specific to IPython.
I also want to use it for sage-run and sagenb and perhaps other places
too which don't use IPython.
--
You received this message because you are subscribed to th
IMHO this would better be solved upstream.
* IPython input transformers should also keep a map from the original
source lines to the transformed lines. By default, that would just be the
identity map.
* IPython already hooks into the linecache used for tracebacks in
get_ipython().compile.cache
14 matches
Mail list logo
