Wireshark-dev: Re: [Wireshark-dev] Viewing code in Gerrit
Date Index · Thread Index · Other Months · All Mailing Lists
Date Prev · Date Next · Thread Prev · Thread Next
From: Gerald Combs <gerald@xxxxxxxxxxxxx>
Date: Mon, 31 Mar 2014 09:53:53 -0700
On 3/24/14 9:03 AM, Evan Huus wrote:
> On Mon, Mar 24, 2014 at 11:26 AM, Bill Meier <wmeier@xxxxxxxxxxx> wrote:
>> On 3/24/2014 9:21 AM, Evan Huus wrote:
>>>
>>>
>>> In summary: the diff is computed locally in javascript, and seems to
>>> be worse than O(n) on the size of the underlying file; viewing the
>>> diff for any file >1k lines may be slow, but if you just let it run it
>>> will finish eventually. Avoid epan/enterprise-numbers and other files
>>> with 100k+ lines, they take many minutes to finish.
>>>
>>
>> IOW: The 2 versions being diff'd are downloaded in total before doing a
>> local compare ?
> 
> It seems so; however it doesn't appear to be the bandwidth that's the
> problem, but the actual diff algorithm. The bottleneck is definitely
> CPU.

Weird. The server has "diff" and "diff_intraline" caches which are used
to "accelerate display of intraline differences when viewing a file"
according to the documentation so *something* is happening on that end.
The default cache size was 10 MiB for each. I increased it to 100 MiB in
case it helps.

BTW, diffs are much faster using the "New Screen" view, at least on my
machine + browser.