can babel results show time of evaluation?

can babel results show time of evaluation?

[Thomas Alexander Gerds]

hi

I would like to overlay the hash info in results lines of this type

#+RESULTS[6e22789c4866716ffb3d0966cbb07c80cacf2d3c]:
[1] 1

by a description which shows the evaluation time, e.g. like this

#+RESULTS[[6e22789c4866716ffb3d0966cbb07c80cacf2d3c][ evaluated 2013-01-28 Mon 09:33]]:

as far as I can see it would require to modify one line in ob-core.el 

(when hash (concat "["hash"]"))

e.g. by some polished version of this hack:

(when hash (concat "[["hash"][" (with-temp-buffer (org-insert-time-stamp (current-time) 'hm)) "]]"))

but, not sure if that is a good idea and what other changes this would require.

best Thomas 

--
sent from Nil

Re: can babel results show time of evaluation?

[Bastien]

Hi Thomas,

Thomas Alexander Gerds <tag@biostat.ku.dk> writes:

> (when hash (concat "[["hash"][" (with-temp-buffer
> (org-insert-time-stamp (current-time) 'hm)) "]]"))
>
> but, not sure if that is a good idea and what other changes this
> would require.

I'm not sure too (both whether this is a good idea and whether 
this would trigger other changes)... but I'd say this is a very 
specific need, and it can be fulfilled by asking the code to
insert a timestamp into the exported results, right?

See for e.g.:

#+BEGIN_SRC emacs-lisp :results raw
(message ";; %s" (format-time-string (cdr org-time-stamp-custom-formats)))
#+END_SRC

HTH,

-- 
 Bastien

Re: can babel results show time of evaluation?

[Thomas Alexander Gerds]

Hi Bastien,

thanks for you reply, in particular for showing how to avoid
org-insert-time-stamp to convert the current-time into a string.

I should have said more clearly what I had in mind with this. I was
thinking of a reproducible report or manuscript with R source blocks
that produce some results. Here org babel allows one to enable caching
for some or all source-blocks. My proposal has two advantages:

1) when visiting the cached results of a particular block after some
time, e.g. one year later, it would be visible *when* the block was
evaluated. This can be useful, e.g., when there are several versions of
the same dataset which are used to produce the cached results.

2) sometimes it may be unclear which blocks are evaluated when running
org-babel-execute-buffer or org-export-as-x, in particular for org-babel
novices. Thus, showing the evaluation time in the results header will
help to learn how caching works.

Cheers
Thomas






Bastien <bzg@altern.org> writes:

> Hi Thomas,
>
> Thomas Alexander Gerds <tag@biostat.ku.dk> writes:
>
>> (when hash (concat "[["hash"][" (with-temp-buffer
>> (org-insert-time-stamp (current-time) 'hm)) "]]"))
>> but, not sure if that is a good idea and what other changes this
>> would require.
>
> I'm not sure too (both whether this is a good idea and whether this
> would trigger other changes)... but I'd say this is a very specific
> need, and it can be fulfilled by asking the code to insert a timestamp
> into the exported results, right?
>
> See for e.g.:
> #+BEGIN_SRC emacs-lisp :results raw
> (message ";; %s" (format-time-string (cdr org-time-stamp-custom-formats)))
> #+END_SRC
> HTH,
--

Re: can babel results show time of evaluation?

[Eric Schulte]

Hi Thomas,

I've just pushed this functionality up to the main branch.  To make use
of it add the following to your Emacs config.

  (setq org-babel-hash-show-time t)

It works for me in some simple tests, and I can confirm that with this
variable set to nil, I get no new failures in the Org-mode test suite,
so it shouldn't introduce any problems.

Thanks for suggesting this functionality,

Thomas Alexander Gerds <tag@biostat.ku.dk> writes:

> Hi Bastien,
>
> thanks for you reply, in particular for showing how to avoid
> org-insert-time-stamp to convert the current-time into a string.
>
> I should have said more clearly what I had in mind with this. I was
> thinking of a reproducible report or manuscript with R source blocks
> that produce some results. Here org babel allows one to enable caching
> for some or all source-blocks. My proposal has two advantages:
>
> 1) when visiting the cached results of a particular block after some
> time, e.g. one year later, it would be visible *when* the block was
> evaluated. This can be useful, e.g., when there are several versions of
> the same dataset which are used to produce the cached results.
>
> 2) sometimes it may be unclear which blocks are evaluated when running
> org-babel-execute-buffer or org-export-as-x, in particular for org-babel
> novices. Thus, showing the evaluation time in the results header will
> help to learn how caching works.
>
> Cheers
> Thomas
>
>
>
>
>
>
> Bastien <bzg@altern.org> writes:
>
>> Hi Thomas,
>>
>> Thomas Alexander Gerds <tag@biostat.ku.dk> writes:
>>
>>> (when hash (concat "[["hash"][" (with-temp-buffer
>>> (org-insert-time-stamp (current-time) 'hm)) "]]"))
>>> but, not sure if that is a good idea and what other changes this
>>> would require.
>>
>> I'm not sure too (both whether this is a good idea and whether this
>> would trigger other changes)... but I'd say this is a very specific
>> need, and it can be fulfilled by asking the code to insert a timestamp
>> into the exported results, right?
>>
>> See for e.g.:
>> #+BEGIN_SRC emacs-lisp :results raw
>> (message ";; %s" (format-time-string (cdr org-time-stamp-custom-formats)))
>> #+END_SRC
>> HTH,
> --
>

-- 
Eric Schulte
http://cs.unm.edu/~eschulte

Re: can babel results show time of evaluation?

[Bastien]

Hi Eric,

Eric Schulte <schulte.eric@gmail.com> writes:

> I've just pushed this functionality up to the main branch.  To make use
> of it add the following to your Emacs config.
>
>   (setq org-babel-hash-show-time t)

Thanks!

-- 
 Bastien

Re: can babel results show time of evaluation?

[Bastien]

Hi Eric,

Eric Schulte <schulte.eric@gmail.com> writes:

> I've just pushed this functionality up to the main branch.  To make use
> of it add the following to your Emacs config.
>
>   (setq org-babel-hash-show-time t)

Well, the patch breaks master, because of (defvar org-ts-regexp).

Can you fix this?

Thanks,

-- 
 Bastien

Re: can babel results show time of evaluation?

[Thomas Alexander Gerds]

Hi Eric

yes, it is in line 178 the call to

(defcustom org-babel-load-languages '((emacs-lisp . t)) ...

which seems to need `org-ts-regexp'

but, otherwise it works! Thank you very much for this option. 

Best
Thomas




Bastien <bzg@altern.org> writes:

> Hi Eric,
>
> Eric Schulte <schulte.eric@gmail.com> writes:
>
>> I've just pushed this functionality up to the main branch.  To make
>> use of it add the following to your Emacs config.
>> (setq org-babel-hash-show-time t)
>
> Well, the patch breaks master, because of (defvar org-ts-regexp).
>
> Can you fix this?
>
> Thanks,
--
Thomas A. Gerds -- Assoc. Prof. Department of Biostatistics
University of Copenhagen, Øster Farimagsgade 5, 1014 Copenhagen, Denmark
Office: CSS-15.2.07 (Gamle Kommunehospital)
tel: 35327914 (sec: 35327901) 

Re: can babel results show time of evaluation?

[Nicolas Goaziou]

Hello,

Eric Schulte <schulte.eric@gmail.com> writes:

> I've just pushed this functionality up to the main branch.  To make use
> of it add the following to your Emacs config.
>
>   (setq org-babel-hash-show-time t)
>
> It works for me in some simple tests, and I can confirm that with this
> variable set to nil, I get no new failures in the Org-mode test suite,
> so it shouldn't introduce any problems.

Well, it breaks the parser, somehow. Indeed, it expects from an
affiliated keyword such as #+results: (or #+caption for that matter) to
have only one secondary value. Sexp representation of these keywords is
already very complicated, it will not help to add values ad nauseum.

Also, if you use Org time-stamp syntax, at some point, someone will ask
the parser to be able to parse it. And it won't.

Eventually, it doesn't help to have hash syntax match Org links. This is
confusing.

Therefore, if this feature really has to be introduced, I suggest, at
the very least, to avoid using Org regular syntax for the reference to
time. It would also help to blur distinction between the time and the
hash string. Perhaps something like the following:

  #+results[2013-01-30T20:15:37 hash-string]: name

Thank you.


Regards,

-- 
Nicolas Goaziou

Re: can babel results show time of evaluation?

[Eric Schulte]

Nicolas Goaziou <n.goaziou@gmail.com> writes:

> Hello,
>
> Eric Schulte <schulte.eric@gmail.com> writes:
>
>> I've just pushed this functionality up to the main branch.  To make use
>> of it add the following to your Emacs config.
>>
>>   (setq org-babel-hash-show-time t)
>>
>> It works for me in some simple tests, and I can confirm that with this
>> variable set to nil, I get no new failures in the Org-mode test suite,
>> so it shouldn't introduce any problems.
>
> Well, it breaks the parser, somehow. Indeed, it expects from an
> affiliated keyword such as #+results: (or #+caption for that matter) to
> have only one secondary value. Sexp representation of these keywords is
> already very complicated, it will not help to add values ad nauseum.
>
> Also, if you use Org time-stamp syntax, at some point, someone will ask
> the parser to be able to parse it. And it won't.
>
> Eventually, it doesn't help to have hash syntax match Org links. This is
> confusing.
>
> Therefore, if this feature really has to be introduced, I suggest, at
> the very least, to avoid using Org regular syntax for the reference to
> time. It would also help to blur distinction between the time and the
> hash string. Perhaps something like the following:
>
>   #+results[2013-01-30T20:15:37 hash-string]: name
>
> Thank you.
>

Are there instances aside from "#+results" where one would want to pack
ancillary information into a keyword?

Perhaps both the hash and the time should be moved *outside* the
keyword, e.g. something like the following,

#+Results: name time hash

What do you think?

>
>
> Regards,

-- 
Eric Schulte
http://cs.unm.edu/~eschulte

Re: can babel results show time of evaluation?

[Eric Schulte]

Bastien <bzg@altern.org> writes:

> Hi Eric,
>
> Eric Schulte <schulte.eric@gmail.com> writes:
>
>> I've just pushed this functionality up to the main branch.  To make use
>> of it add the following to your Emacs config.
>>
>>   (setq org-babel-hash-show-time t)
>
> Well, the patch breaks master, because of (defvar org-ts-regexp).
>

Oh, my apologies.  So this is a defconst not a defvar.

>
> Can you fix this?
>

As a temporary fix, I put the full defconst invocation into ob-core.el,
but this can't be the best solution.

What's the "best practice" for how to use a defconst defined in an
external file (when you can't require that file because that file
requires you)?

Thanks,

-- 
Eric Schulte
http://cs.unm.edu/~eschulte

Re: can babel results show time of evaluation?

[Achim Gratz]

Eric Schulte writes:
> What's the "best practice" for how to use a defconst defined in an
> external file (when you can't require that file because that file
> requires you)?

Putting all definitions into a separate "leaf" file that can then be
required from everywhere with impunity.  Bastien doesn't like this idea,
so the only way out would be autoloading such symbols, but of course
autoloading a defconst is just a glorified way to put its definition in
a leaf file.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Factory and User Sound Singles for Waldorf rackAttack:
http://Synth.Stromeko.net/Downloads.html#WaldorfSounds

Re: can babel results show time of evaluation?

[Nicolas Goaziou]

Eric Schulte <schulte.eric@gmail.com> writes:

> Are there instances aside from "#+results" where one would want to
> pack ancillary information into a keyword?

So far, only #+results and #+caption (for the short caption) are
concerned. See `org-element-dual-keywords' for some more details.

> Perhaps both the hash and the time should be moved *outside* the
> keyword, e.g. something like the following,
>
> #+Results: name time hash
>
> What do you think?

I think the hash string is fine where it is since there's already
a keyword category for such constructs. It's even handy in the case of
captions. It would be better if we don't overload this location, though.

As far as the parser goes, both

    #+results[mix-time-and-hash]: name

and

    #+results[hash]: name time

are fine, as long as time isn't, as already explained, exactly an Org
timestamp. The latter is perhaps more elegant. I guess it depends on how
important the time information is. Anyway, use whichever suits you.


Regards,

-- 
Nicolas Goaziou

Re: can babel results show time of evaluation?

[Bastien]

Hi Eric,

Eric Schulte <schulte.eric@gmail.com> writes:

> As a temporary fix, I put the full defconst invocation into ob-core.el,
> but this can't be the best solution.

Yep, having the same defconst twice...

> What's the "best practice" for how to use a defconst defined in an
> external file (when you can't require that file because that file
> requires you)?

To circumvent the dependency problem by having defconst in a separate
file that any file can load safely -- since that file will not require
any other file.

I'm not ready yet to take that road for now, because it requires time
and attention.  I just hardcoded the regexp in ob-core.el and added 
a warning.

Thanks,

-- 
 Bastien

Re: can babel results show time of evaluation?

[Bastien]

Hi Achim,

Achim Gratz <Stromeko@nexgo.de> writes:

> Putting all definitions into a separate "leaf" file that can then be
> required from everywhere with impunity.  Bastien doesn't like this
> idea,

Maybe you misunderstood me.

I am very much in love with the idea of putting defconst in a
separate file.

What I don't like *at all* is the idea you proposed to put every
variable (including user options) in a separate file.  The reason why
I don't like this is because it will force potential contributors to
modify this "Yeah-I-contain-all-variables!"-file.  This would be a
maintainance burden.

-- 
 Bastien