Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
-
this.title = "gg Microsoft"
this.metadata = {
rant: true,
long: true,
super_long: true,
has_summary: true
}
// Also:
let microsoft = "dead" // please?
tl;dr: Windows' MAX_PATH is the devil, and it basically does not allow you to copy files with paths that exceed this length. No matter what. Even with official fixes and workarounds.
Long story:
So, I haven't had actual gainful employ in quite awhile. I've been earning just enough to get behind on bills and go without all but basic groceries. Because of this, our electronics have been ... in need of upgrading for quite awhile. In particular, we've needed new drives. (We've been down a server for two years now because its drive died!)
Anyway, I originally bought my external drive just for backup, but due to the above, I eventually began using it for everyday things. including Steam. over USB. Terrible, right? So, I decided to mount it as an internal drive to lower the read/write times. Finding SATA cables was difficult, the motherboard's SATA plugs are in a terrible spot, and my tiny case (and 2yo) made everything soo much worse. It was a miserable experience, but I finally got it installed.
However! It turns out the Seagate external drives use some custom drive header, or custom driver to access the drive, so Windows couldn't read the bare drive. ffs. So, I took it out again (joy) and put it back in the enclosure, and began copying the files off.
The drive I'm copying it to is smaller, so I enabled compression to allow storing a bit more of the data, and excluded a couple of directories so I could copy those elsewhere. I (barely) managed to fit everything with some pretty tight shuffling.
but. that external drive is connected via USB, remember? and for some reason, even over USB3, I was only getting ~20mb/s transfer rate, so the process took 20some hours! In the interim, I worked on some projects, watched netflix, etc., then locked my computer, and went to bed. (I also made sure to turn my monitors and keyboard light off so it wouldn't be enticing to my 2yo.) Cue dramatic music ~
Come morning, I go to check on the progress... and find that the computer is off! What the hell! I turn it on and check the logs... and found that it lost power around 9:16am. aslkjdfhaslkjashdasfjhasd. My 2yo had apparently been playing with the power strip and its enticing glowing red on/off switch. So. It didn't finish copying.
aslkjdfhaslkjashdasfjhasd x2
Anyway, finding the missing files was easy, but what about any that didn't finish? Filesizes don't match, so writing a script to check doesn't work. and using a visual utility like windirstat won't work either because of the excluded folders. Friggin' hell.
Also -- and rather the point of this rant:
It turns out that some of the files (70 in total, as I eventually found out) have paths exceeding Windows' MAX_PATH length (260 chars). So I couldn't copy those.
After some research, I learned that there's a Microsoft hotfix that patches this specific issue! for my specific version! woo! It's like. totally perfect. So, I installed that, restarted as per its wishes... tried again (via both drag and `copy`)... and Lo! It did not work.
After installing the hotfix. to fix this specific issue. on my specific os. the issue remained. gg Microsoft?
Further research.
I then learned (well, learned more about) the unicode path prefix `\\?\`, which bypasses Windows kernel's path parsing, and passes the path directly to ntfslib, thereby indirectly allowing ~32k path lengths. I tried this with the native `copy` command; no luck. I tried this with `robocopy` and cygwin's `cp`; they likewise failed. I tried it with cygwin's `rsync`, but it sees `\\?\` as denoting a remote path, and therefore fails.
However, `dir \\?\C:\` works just fine?
So, apparently, Microsoft's own workaround for long pathnames doesn't work with its own utilities. unless the paths are shorter than MAX_PATH? gg Microsoft.
At this point, I was sorely tempted to write my own copy utility that calls the internal Windows APIs that support unicode paths. but as I lack a C compiler, and haven't coded in C in like 15 years, I figured I'd try a few last desperate ideas first.
For the hell of it, I tried making an archive of the offending files with winRAR. Unsurprisingly, it failed to access the files.
... and for completeness's sake -- mostly to say I tried it -- I did the same with 7zip. I took one of the offending files and made a 7z archive of it in the destination folder -- and, much to my surprise, it worked perfectly! I could even extract the file! Hell, I could even work with paths >340 characters!
So... I'm going through all of the 70 missing files and copying them. with 7zip. because it's the only bloody thing that works. ffs
Third-party utilities work better than Microsoft's official fixes. gg.
...
On a related note, I totally feel like that person from http://xkcd.com/763 right now ;;21 -
Got to a client, we are taking over their software dev and IT.
"you're not touching my code ". Listen fuck twit, a robocopy script is not fucking code.
"I wrote a 3500 line code for this company" no fuck twit, you wrote a whole lot of fucking gibberish that looks like someone shat out BASH and it met html along the way. It doesn't compile, it doesn't run, it's a fucking dormant file. You charged people for shit all.
Setting up exchange is also not a big whoop.
Moving them over to CentOS server (he had them on XP still), and writing enough code to qualm my frustration at people.4 -
*wants to download some YouTube videos in youtube-dl*
$ youtube-dl --a-bunch-of-options
> Can't download this, sorry.
*realizes that Ubuntu probably has an outdated version like usual*
# apt remove youtube-dl
*Realizes that this steaming pile of shit pulled in some icons and Wayland on a headless server*
# apt autoremove
> 300-something MB cleared
For a command-line tool...
# pip install youtube-dl
# apt install ffmpeg
$ youtube-dl --audio-format mp3 -o "%(title)s.%(ext)s" https://youtube.com/playlist/...
> Sorry mate, a video was removed from this playlist! Let me go ahead and shit the bed on this issue that's been reported several times since 6 years ago.
*finds an issue on GitHub reporting this, add -i option to continue on error*
$ youtube-dl --audio-format mp3 -o "%(title)s.%(ext)s" https://youtube.com/playlist/... -i
> There you go, your .webm files as requested!
But.. I requested .mp3 output? --audio-format mp3, don't you see?
> Oh no you need to add in another option to tell me to actually do that first. --extract-audio, you see?
But why.. why do you need to be told that twice? Oh ffs, fuck it.
Reminds me of robocopy. That shit required me to tell it 25 times what to do and it'd still not do it right as well. And you know what, compared to rsync where -avz works 99% of the time, I hate it.7 -
Back when I migrated my file server to a VM, I used robocopy to copy everything over (and caused quite a few fuckups before it'd behave halfway decent.. or so I thought). Now I tried to add my desktop's SSH key to the authorized_keys on said server but it wouldn't accept my key after that. After a bit of digging I've found that my entire home directory (where the file server hosts its mirror of my D: drive) had everything set to 777, and that's why the key isn't accepted. Great permission mode, isn't it? Much secure, very wow! Thank you so much robocopy!!!3
-
Many of you who have a Windows computer may be familiar with robocopy, xcopy, or move.
These functions? Programs? Whatever they may be, were interesting to me because they were the first things that got me really into batch scripting in the first place.
What was really interesting to me was how I could run multiples of these scripts at a time.
<storytime>
It was warm Spring day in the year of 2007, and my Science teacher at the time needed a way to get files from the school computer to the hard-drive faster. The amount of time that the computer was suggesting was 2 hours. Far too long for her. I told her I’d build her something that could work faster than that. And so started the program would take up more of my time than the AI I had created back in 2009.
</storytime>
This program would scan the entirety of the computer's file system, and create an xcopy batch file for each of these directories. After parsing these files, it would then run all the batch files at once. Multithreading as it were? Looking back on it, the throughput probably wasn't any better than the default copying program windows already had, but the amount of time that it took was less. Instead of 2 hours to finish the task it took 45 minutes. My thought for justifying this program was that; instead of giving one man to do paperwork split the paperwork among many men. So, while a large file is being copied, many smaller files could be copied during that time.
After that day I really couldn't keep my hands off this program. As my knowledge of programming increased, so did my likelihood of editing a piece of the code in this program.
The surmountable amount of updates that this program has gone through is amazing. At version 6.25 it now sits as a standalone batch file. It used to consist of 6 files and however many xcopy batch files that it created for the file migration, now it's just 1 file and dirt simple to run, (well front-end, anyways, the back-end is a masterpiece of weirdness, honestly) it automates adding all the necessary directories and files. Oh, and the name is Latin for Imitate, figured it's a reasonable name for a copying program.
I was 14, so my creativity lacked in the naming department >_<
1 -
I just gave robocopy another try, in order to get my WanBLowS D: drive and my file server synchronized again, in preparation to move that file server VM to a LXC container instead.. bad choice. I should've used rsync in WSL.
Hey you Not so Robust File Copier for WanBLowS, how many attempts of you fucking up my file server's dotfiles does it take before I configure you right with every fucking option you have specified? How about you actually behave somewhat decently like rsync where -avz works 99% of the time, in local, remote, any scenarios that you can think of that aren't super obscure?! HOW DIFFICULT CAN IT BE, REDMOND CERTIFIED ENGANEERS?!!
Drown in a pond of bleach, Microshit certified MOTHERFUCKERS!!!!
Well, at least this time it didn't fuck up my .ssh directory so I can still authenticate to the VM.. so I guess that at least that's a win. Even that you can't take for granted anymore with this piece of garbage!!!4 -
WanBLowS, all I ask you, the only thing I ask you to do now, is to synchronize some files from A to B without transferring the whole goddamn 1.3TB of stuff that for the most part hasn't changed in any way, other than whatever your crappy NTFS filesystem mutated it into.
Robocopy, rsync, even Windows' built-in explorer. None of them do the job as they should. Why Windows.. why?! Why can't you just do one thing properly for once?!!! Piece of junk!17 -
Just now when I'm watching one of the many anime's I've saved onto my file server I noticed something.. all of their files are incomplete, and so are they on the NTFS mirror on this WanBLowS host. The files got corrupted. I recall that I used robocopy to place the files back and forth, and yet again it lives up to its expectations of it being a motherfucking piece of Winshit. FUCK YOU ROBOCOPY!!! If I wanted to fetch that anime yet again just to deal with your developers' incompetence, I'd have watched it online!! Meanwhile tell me, HOW DIFFICULT IS IT TO DEAL WITH A NETWORK FILE TRANSFER THAT EVEN USES YOUR OWN SHITFEST OF A PROTOCOL, FUCKING SMB?!! MSFT certified pieces of shit!!!!7
-
This was more than 15 years ago. We migrated a bunch if data (home to a new server and repurposed the old one, the same night. This was not the first task on that allweekender, so it was around 3am on Sunday, with very little sleep, when I had to copy the data. I did that by logging in as admin and copying with Total Commander. Obviously, even admin did not have the permissions to some folders, so a lot of financial data were lost, as the users found out on Monday morning. We had no backup. Old server was not only reformatted, but the disks were used to build a different raid set. Luckily, one of the users who had access to this data kept a backup on a flash drive. (If you're wondering, I should've used robocopy with backup mode)
-
When you have to delete/move 1000-character files that Dropbox synced from a Mac OS source...
So thankful for `robocopy` -- even though you'd think MS would eventually allow longer filenames.
Top Tags
Weekly Rant
View