Re: rewriting history

[<pl...@pi...>]

On 11/19/13 09:17, Daniel J Sebald wrote:
> On 11/19/2013 01:03 AM, pl...@pi... wrote:
>> On 11/19/13 07:05, Daniel J Sebald wrote:
>>> On 11/18/2013 02:58 PM, pl...@pi... wrote:
>>>> Hi,
>>>>
>>>> is there was way to stop gnuplot compacting history?
>>>
>>> Do you mean that the current command is not repeated in history if it
>>> matches the previous most recent command in the history buffer?
>>>
>>> My history looks pretty lengthy, but I do seem to recall a limit. Also,
>>> given it isn't recent history I can't look through it and conclude that
>>> it is complete.
>>
>> yes, it's a feature that dupes are eliminated. Just try repeating your
>> last plot command a few times. Only the last occurrence will appear in
>> history.
>>
>> I've been pulling my hair trying to work out why a fit command is
>> producing different results. Having an accurate record of my command
>> history would be an immense help.
>>
>> I'm currently having to cut and paste every command into a text file as
>> I go along which is a serious PITA.
>
> That's sort of my modus operandi, except I edit in a file and either
> select and copy with the center mouse button or save and load the file.
>   History is nice for tweaking commands and trial-and-error, but
> anything more than five commands deep is tough to edit.  Nonetheless,
> history modifications might be worthwhile.
>
> Dan
>
>
>>
>> Peter.
>>
>>
>>
>>>
>>>
>>>> While I realise this is done to save storage space somewhere, this is a
>>>> constant annoyance.
>>>>
>>>> I have an annoying problem where something keeps changing and I need to
>>>> track exactly what is happening.
>>>>
>>>> In a shell I would use the history command to see _exactly_ what I did
>>>> and in what order to see what I have changed. I can see a true
>>>> record of
>>>> my commands and repeat various parts of it until I reproduce and
>>>> understand the error.
>>>>
>>>> with gnuplot this does not seem possible because history is forever
>>>> being revised and compacted and no longer retains a useful record. In
>>>> fact "history" does not provide the command history. It's more like
>>>> recent favourites, that often saves some typing but it is not the
>>>> command history.
>>>>
>>>> I'm not that tight for space that I need to economise a few K. I'd much
>>>> rather save the few hours it's going to take me to find this problem
>>>> without a proper record of the commands I've issued to make the problem
>>>> repeatable.
>>>>
>>>> Is there a means of preventing gnuplot from compacting the command
>>>> history be removing dupes?
>>>
>>> I recall some years ago writing a history patch that allowed it to be
>>> configured in a couple ways: one being complete history, one being more
>>> condensed. I might still have that around if you are interested, but
>>> I'd have to search for it and I'm sure it has conflicting hunks by now.
>>>
>>> Dan
>>>
>>
>>
>

For this to be of most use to me I would like to have (at least options 
on ) recording verbatim command history and history display without 
numbers, which would make copying to a runnable script file less laborious.

m2c, Peter.