Better Time Estimations

[Void]

Just wanted to mention that it would be really great if the time estimations could be improved. They're bad enough in normal everyday use, leading to many a time of family waiting upset for me to get out as I only wait for a copy to finish and it lasts many minutes past the estimated time even after 50% progress, but now I'm actually doing a big hdd move and i've seen quite a couple of time seconds that last for many minutes. For example in the last 5 or more minutes my estimation just changed from "25seconds remaining" to "21 seconds remaining". If you could do something about this it would be really cool.

I'm guessing the problem might come from estimating just sizes, and maybe smaller file sizes mess up the estimation. Isn't there some way that could be compensated. Or even better (possibly simpler?) somehow to readjust the estimation of time based on the success of the previous estimation/how much time has passed by percentage of copy? I mean when a copy/move that was estimated as at a few seconds lasts more than 10 minutes for n files, maybe one could assume that also the next n files will take another 10 minutes?

Of course all of this I'm asking very humbly and thankfully. I and my wife are so very thankful for the great work you've put into MC and for giving it out for free. It has taken up a major part in our daily computer use. Thank you!

[Mathias (Author)]

Estimate time left is just a estimate. It can never be perfect because so many external factor affect how long it takes and .  Right now estimated time is calculated, based on how long time it has taken to copy the bytes up to that point, and how many bytes that is left.

But if you copy a lot of small file the harddrive will have a lot of overhead when it create and closes files. and also drive fragmentation and other program on the computer that also starts to access the drives will affect the performance.

[Void]

I'm assuming before the copy a list is compiled of the fieles to be copied (maybe including their file sizes?), couldn't this somewho be used to create an average file size+time of copy average so that if one copies say 10 big files and 10000 small files of the same physical size at the end the last files wouldn't be estimated to take just as much as the first half?

[Mathias (Author)]

I'm assuming before the copy a list is compiled of the fieles to be copied (maybe including their file sizes?), couldn't this somewho be used to create an average file size+time of copy average so that if one copies say 10 big files and 10000 small files of the same physical size at the end the last files wouldn't be estimated to take just as much as the first half?

It is. And the time estimate is based on bytes copied. if we have 100G to copy and we have copied 50G in 5Min. then the estimate is that the other 50G would also take 5 min. Problem is that if the first 70G are very large files and the last 30G are 3.500.000 very small files.  Then the estimates will be off since the overhead of creating files. File systems are slow when adding/removing entire into the filesystem.

[Void]

My question would be if the program learns as it's copying about the system it's working on so as to improve the estimation. So in your example if the program started copying those 3.500.000 small files you mentioned, after 2 million of them each taking a huge amount of overheard, shouldn't the program start to figure out by now that the next million of similarly small sized files might take also a long time, as opposed to sticking to it's optimistic initial estimate of 25 minutes ago? More than one time this for me this was tens of minutes when every estimated second took minutes to pass, yet I could see no improvement in the quality of the estimation of the final time. Wouldn't it be cool if the program learned a bit/reacted to what it was copying as it was copying it? MC could be a pioneer in the field where most annoying explorer interfaces & installers are legendarily funny for spending another half the time between 99% and 100% marks of a process

[Mathias (Author)]

That would be too complex and time consuming to develop something like that. If it is would even be possible.

It would be an interesting research project for a student or similar but way to time consuming for me to work on.

[Void]

I understand and respect the reasoning! Thanks! Hope it's okay that I asked though.

[Mathias (Author)]

Sure, no problem