[Python-mode] pdbtrack

[Andreas Roehler]

ken manheimer wrote:
> looking at this a little bit more closely, it's surprising to me not
> that it fails for you, but that it works at all.  


Hi Ken,

thats the point probably.

Originally question was: must we change the source-code in order to
make pdbtrack working, must we put a pdb.set_trace() into it?

Tried to avoid that and watched results coming from pdb also without that.

While executing, results are delivered to function

py-pdbtrack-get-source-buffer (block)

line 1444 of my python-mode.el


Then (string-match py-pdbtrack-stack-entry-regexp block) decides to accept result or not.

Played a little bit with py-pdbtrack-stack-entry-regexp:.

;; pdbtrack constants
(defconst py-pdbtrack-stack-entry-regexp
;  "^> \\([^(]+\\)(\\([0-9]+\\))\\([?a-zA-Z0-9_]+\\)()"
;; "^> \\(.*\\)(\\([0-9]+\\))\\([?a-zA-Z0-9_]+\\)()"
  "> \\(.*\\)(\\([0-9]+\\))\\(.*\\)()"
  ;; "^ \\(.*\\)"
"Regular expression pdbtrack uses to find a stack trace entry.")



Andreas

have you made any
> changes to the py-pdbtrack-input-prompt variable (aka
> 'python-pdbtrack-input-prompt' in recent versions of python.el)?
> 
> as the docstring for py-pdbtrack-track-stack-file states, "We depend on
> the pdb input prompt matching `py-pdbtrack-input-prompt' at the
> beginning of the line."  the default setting simply should not match the
> "ipydb> " that your photos are showing.  ?
> 
> sorry if i'm missing something obvious - i'm trying to get context here,
> with not quite enough time to pay full attention...
> -- 
> ken
> http://myriadicity.net <http://myriadicity.net/>
> 
> On Fri, May 1, 2009 at 5:28 PM, ken manheimer <ken.manheimer at gmail.com
> <mailto:ken.manheimer at gmail.com>> wrote:
> 
>     On Fri, May 1, 2009 at 4:50 PM, Andreas Roehler
>     <andreas.roehler at online.de <mailto:andreas.roehler at online.de>> wrote:
> 
>         Barry Warsaw wrote:
>         > On Apr 30, 2009, at 4:07 PM, Andreas Roehler wrote:
>         >
>         >> I'll send you two screenshots offlist.  Please feel
>         >> free to forward them to interested persons, just didn't
>         >> want to publish my path at the list.
>         >>
>         >> 20090428_pdbtrack3.png displays pdbtrack opened second
>         >> windows, cursor displayed at line 4 "import"
>         >>
>         >> With 20090428_pdbtrack4.png you see  shell-output
>         >> from line 8, but cursor in second window still is at line 4.
>         >>
>         >> Always get "Traceback cue not found"
>         >>
>         >> Cause seems var `py-pdbtrack-stack-entry-regexp'.
>         >>
>         >> That doesn't happen, if pdb.set_trace() is inside the
>         >> python-file.
>         >>
>         >> Did someone else remark this?
>         >
>         > Hi Andreas,
>         >
>         > I haven't heard this one before, but perhaps it's because of
>         the ipython
>         > prompt?
> 
>         Maybe. As shows message in first screenshot #3, pdbtrack
>         indicates line 4 correctly.
>         Its not brocken completely.
> 
> 
>     i'm having some difficulties tracking your descriptions of the
>     situation, andreas, but i think you (and barry) are right that the
>     problem is failure to recognize, rather than pdbtrack not being
>     activated (as you seemed to initially be suggesting).
>      
> 
>         Its a question of regexp, which doesn't recognise/accept the
>         result, delivered by `block'.
> 
> 
>     i don't understand what you mean by "delivered by 'block'"
>      
> 
>         Already tried to change the regexp or even simply eliminate that
>         checking step, but not
>         found a solution.
> 
>      
>     the match may be necessary to parse the traceback, and for other
>     reasons.
> 
>         Hhm. Could you give me an example, how you run script activating
>         pdbtrace, reaching
>         the standard (pdb) prompt from Emacs?
> 
> 
>     what do you mean by "reaching the standard (pdb) prompt from
>     Emacs"??  perhaps you're talking about the python interaction buffer
>     that python-mode provides?  i believe, but am not certain, that
>     pdbtrack watches the output in any comint buffer, including the
>     python interaction buffer.  i can check this kind of thing, but need
>     to understand better what you're saying before investigating.
> 
>         Only get it from shell.
> 
> 
>     one thing i can suggest that you do is copy the text from the emacs
>     buffer that pdbtrack should recognize but doesn't, and send that
>     excerpt to us.  i know that's in the pictures, but it's easier for
>     you to send us the text than it is for us to type it accurately from
>     the pictures.-)  preferably you would also send an excerpt that is
>     similar but is successfuly recognized.
> 
>     i will try to find time to examine it, and see if i can identify
>     what is happening and what should be happening...
>     -- 
>     ken
>     http://myriadicity.net
>      
> 
>         Thanks
> 
>         Andreas
> 
> 
>          I think pdbtrack expects (by default) the standard (pdb) prompt.
>         >
>         > -Barry
>