Search, Replace and Unicode

[Phil Davis]

On 12/29/12 11:52 AM, Richmond wrote:
> On 12/29/2012 09:38 PM, Phil Davis wrote:
>> Hi Richmond,
>>
>> Just curious - does setting the lockMessages & lockScreen to true 
>> speed it up any? Or hiding the field? (Sorry if these have already 
>> been answered; I'm not following the thread too closely)
>
> I really don't know as I am so "thick" those possibilities had not 
> occurred to me.
>
> However, I cracked open my monster Sanskrit font and looked at it 
> FontForge and realised that I was being fairly bl**dy silly crunching 
> through
> thousands of irrelevant unicode addresses, so chopped things up into 3 
> REPEAT UNTIL loops for the unicode ranges that were relevant to my work:
>
> 2325 - 2431
> 57354 - 58498
> 61952 - 62738
>
> giving a saving of some 58371 interations!!!! and boiling the whole 
> thing down to about 5 seconds . . . that is on the same sample text I
> previously used.

Wow! Nice gain.

>
> HOWEVER . . . now I will play around with setting lockMessages and 
> lockScreen as per your suggestions to see if there is any appreciable 
> gain.
>
> Phil; you state that you are "not following the thread too closely"; 
> is that because it is not something that interests you, or do you have 
> a stake in Unicode text manipulation but have been busy elsewhere.

I am being dragged toward Unicode, kicking and screaming. It looms large 
in my future, as one client has asked me to add support for Arabic in 
his training system soon. But I don't have much experience with it. I'm 
hoping the improvements in LC 5.5 and beyond will make for a little 
smoother landing in UnicodeLand.

>
> I sometimes get the feeling (err . . . paranoid) that I am the only 
> person using RR Livecode who is doing 'serious sh*t' with unicodeText;
>
> but, hey, why should I worry?; in 1985, at the University of Durham, I 
> was the only student trying to process text with PASCAL 5 while all the
> Physics students were crunching numbers.
>
>>
>>
>> On 12/29/12 8:18 AM, Richmond wrote:
>>> ran it through unicode addresses from 2325 to 62738, which took 
>>> about 90 seconds, which is rather too long, so
>>> will cut that down to run through several ranges of unicode 
>>> addresses rather than the whole shebang.
>>
>
>
> _______________________________________________
> use-livecode mailing list
> use-livecode at lists.runrev.com
> Please visit this url to subscribe, unsubscribe and manage your 
> subscription preferences:
> http://lists.runrev.com/mailman/listinfo/use-livecode
>

-- 
Phil Davis