zakalibit

As well filepush has no clue if the buffer starts at the start of the packet, so let say you read in to provided buffer 2000 bytes, so the first chunk of data starts at the packet boundary, files push will write only 1880 bytes, i.e. 120 bytes from the last incomplete packet will be lost, next read let say the same 2000 bytes, will not start at the beginning of the packet, filepush again will throw last 120 bytes, so your hardware will throw away the first 68 bytes and the last 120 bytes as you have an incomplete packet at the start and the end, so you start loosing 2 packets per read, in my implementation I make sure that buffer starts at the beginning of the packet and it will keep the last incomplete packet content for the next read, so that filtering is is nicely feeding in to filepush chunks of buffer that have a certain number of full packets, so nothing gets chopped by filepush and nothing filtered away by hardware.

So what are the next steps? obviously at the end the code should adhere to the core members opinion

Would it work this way? I'll apply one by one patch and issue a merge request, so we can review the individual patch, I wont apply the next patch until the current one is reviewed.

@betacentauri thanks again for doing the individual patches

https://sourceforge....ge-requests/22/

Another one, mostly cosmetic, it has some performance improvement aspects, but I do not think that will be noticeable by users, but every little helps

https://sourceforge....ge-requests/21/