bigcanuck

per hub customization

29 posts in this topic

FL will be generated in settings directory when needed (use File >> Open own file list to force it's creation). 40K files would generate around 1,5MB FL?

That would be around correct yes. Though I think with generating multiple filelist system won't be a huge problem. You only need the one has table right. So like you'll have your full filelist (whatever it is) and then you can just have a 2nd filelist which essentially has pointers to the correct hash index. So when a user calls your filelist lets call it filelist B Apex will look at the call and say it's coming from hub B so use list B. Then while using list B it'll simply go. list B contains said item indexs, return items from index in List A( your general list )

Share this post


Link to post
Share on other sites

Why do you even worry about the FL size ? Is 1MB (or 10) such a waste of space for you ? If you don't want second filelist don't use it...You should worry about the tech. side of adding this feature - I guess it'll take much coding and thats way i ain't sure if it'll be implemented at all.

Little explanation about how Apex can hash the files (its like bigcanuck suggested)

1.When a filelist is refreshed it should check is the file thats going to be hashed already in the GHI//General Hash Index//This can be done by comparing the file path...

1a.If its hashed add the hash for it and go to the next file.

1b.If its not hashed - hash it in the GHI and then go to point 1.

There also should be a function to clean up the files in the GHI that ain't used.

Share this post


Link to post
Share on other sites

And more, since later on I discovered how to "generate" my FL (1 162KB), it means that f. ex. 50MB will be used only if 5 users ask simultaneously the 5 different "versions" of your share. And after they Dd it, the space is regained... Or at least we should be able to do it in tat way.

Share this post


Link to post
Share on other sites

5-6MB in Your case. Probably much less then hashindex + hashdata.

FL size isn't a problem here... but the actual implementation of generating and distributing them.

Although I write all this - it's definitely not my top REQ. I wonder why devs keep saying that it's possible only on ADC. Maybe it would simply be easier or they plan to implement it some other, simple way. If so I wouldn't really force them...

Share this post


Link to post
Share on other sites