Ah it's okay then. It's probably on the site anyway. I'll probably find a way to get it. I just like the hi10 realeases, 1080p at reasonable file sizes, even though I think Tatami might be 480p. Iirc JSU uses some hardcoded subtitles sometimes.
If I were to pick it would be Hellsing, Spice and Wolf and Tatami Galaxy. Though it would be nice to have a backup or torrents of all of them. I have 2 shows I could contribute but can only seed 100 KB/up :(. Or maybe hi10 will pull something out of their hats and publish the torrents themselves publicly.
A bit unrelated to AT but does anyone have some of the stuff that hi10anime recently took down due to DMCA? I have some of the series ... we could at least trade.
We also store fonts, as well as image subtitles (VOBsub/PGS), which take up a fair amount of space. For simplicity, these all go through the same system as text subtitles.
I just checked the stats: attachments currently consume 34GB (compressed) of space. Uncompressed, this would be 192GB. So the ratio is a bit higher than your estimate. Note that the attachments system does de-duplicate files, e.g. only ever stores one copy of a font, so there's no efficiency to gain there.
On the server side, compression provides three main benefits: cost savings related to disk space, as well as savings in bandwidth usage, and just easier data handling. For the former, it means 2*158GB less disk consumption (*2 due to the fact that I have a single data replica; I currently don't actually have enough backups of this data, so it may be more in the future), and this figure would increase over time. I don't see bandwidth usage being significant, so won't say much about it. As for the last point, smaller data sets make recovery and setting up a new server quicker.
For visitors, it means faster downloads which consume less bandwidth. It's probably not much for many ASS files, but can be useful for the All Attachments link.
Whilst there's a lot of benefits for me to have files stored compressed, I don't mind serving uncompressed copies of subtitles, but I don't have anything which can do this. If you (or someone else) can write up an nginx module similar to mod_gunzip, but for .xz files, I can install it for serving decompressed subtitles.
You can also try sorting by file type to get the files together. Alternatively, if 7z.exe is available in your PATH environment, you could type something like cmd /k 7z x *.xz into the address bar to extract all .xz files.
As for single files, after downloading, if 7-Zip is associated with .xz files, you should just be able to click on the downloaded item, which opens it in 7-Zip, then double-click the file to open in Agesiub. This saves you from having to navigate to a folder and right-click -> extract.
I get that this does require an extra step somewhere, and thanks for raising that, but it is useful for those on slower and/or metered connections, as it reduces the amount needed to download. I don't mind providing the option for an uncompressed download, however I don't have any code/program which can do this. If you can (or you can get someone else to) write an nginx module similar to mod_gunzip but decompresses .xz files, I can install it to provide uncompressed subtitles.
Is it more accurate? Since it only concerns the the sub file and the rest all the same, its like storageOverTime = (x + w)t changed into storageOverTimeNew = (x + 4w)t, where w is sub file size, t is time, x is the rest like new files and stuff. But w is very small, like less than 100 kb. The extra storage being paid for is unlikely to be in the order of 100's of KBs, right?
Let's say there are 100 files a day and 365 days a year, that's 36500 files a year. So each year, only (25*4KB)*36500 files a year which is 3.65GB more a year.
It would be more accurate to say that someone would have to pay for 4x the storage on an ever increasing database over several years for no compelling reason.
15/08/2017 18:26 — Anonymous