Bug#954845: poor performance for -g 256x50
On 2020-03-24 19:23 -0400, Thomas Dickey wrote:
> On Tue, Mar 24, 2020 at 01:15:40PM +0100, Harald Dunkel wrote:
>> Package: xterm
>> Version: 353-1
>>
>> xterm -g 256x50 has a much worse "scroll performance" than xterm -g 255x50,
>> esp if there are very long lines and its a slow network connection. Sample:
>>
>> xterm -g 255x50:
>> # time cat /var/log/rsnapshot.log
>> :
>> :
>>
>> real 0m0.041s
>> user 0m0.000s
>> sys 0m0.006s
>>
>>
>> xterm -g 256x50:
>> # time cat /var/log/rsnapshot.log
>> :
>> :
>>
>> real 0m27.631s
>> user 0m0.000s
>> sys 0m0.004s
>>
>>
>> Maybe this could be improved?
>
> maybe - but I have to see how to reproduce the problem.
>
> Presumbly it's xterm, but then again, I can see cases where the
> X server (and its uneven quality-of-implementation for certain Xlib calls)
> could explain the problem. If I can reproduce the problem, I'll have no
> doubt :-)
>
> I'm assuming that your example file's kind of large and wouldn't be suitable
> for an attachment (even compressed).
If the file can be cat'ted in 0.05 seconds, it likely is not very big
after all.
> Unless there's something special about
> columns 255/256 -- or multiples of that (such as a multibyte character that has
> to be split), then I'm not sure how to advise on reducing it to some repetitive
> string that could be scripted. But some sort of bisecting approach, making
> successively smaller slices from the original file (and repeating them) might
> work.
>
> How large is the file ("wc" output would help)?
Harald, could you please follow up here?
Cheers,
Sven
Reply to: