Hi, I came across the following issue: using a directory and path reader (filtering for files only) and supplying an empty folder works fine (it doesn't produce any files, but that's just fine). However, say I want to skip any processing from this reader, as an innocent user I supply the Windows nul: device. Now my workspace crashes, how is that? Of course, there are a lot of ways around this problem, but wouldn't it be nice if the directory and path reader behaves in the same way like, say, the Windows dir command?
Page 1 / 1
To answer your question, I'd say "yes it would". But that doesn't help you. It's probably better to submit a support case to Safe. :)
To answer your question, I'd say "yes it would". But that doesn't help you. It's probably better to submit a support case to Safe. :)
Does sound like a bug. I'll ask the team to file it.
Does sound like a bug. I'll ask the team to file it.
Does sound like a bug. I'll ask the team to file it.
ID is C127744
@helmoet I'm trying to reproduce this issue, but I'm not entirely sure how you supplied the Windows NUL: device. Currently, the PATH reader takes any way I try to input the NUL: device and reads it as part of the path for the dataset, and I'm not sure how to convince the reader to accept the NUL: device. How did you supply the NUL: device?
Hi @rileybecker If you want to reproduce, simply place the reader (e.g., take the folder c:\temp) and connect it to a logger, in a new workspace. If you run the workspace it will do just fine. If you prompt and run (SourceDataset_PATH will be published automatically), and supply the value NUL: the following error occurs:
Trying to find a DYNAMIC plugin for reader named `PATH' Loaded module 'PATH' from file 'C:\apps\FME\plugins/PATH.dll' FME API version of module 'PATH' matches current internal version (3.8 20151127) Opening the PATH Reader on folder 'nul:' Using Glob Pattern '*' Allowed Path Type set to 'ANY' Retrieve file properties 'false' The source dataset must be a folder not a file. The dataset specified was 'nul:' A fatal error has occurred. Check the logfile above for details FME Session Duration: 0.1 seconds. (CPU: 0.1s user, 0.0s system) END - ProcessID: 12096, peak process memory usage: 35988 kB, current process memory usage: 35988 kBIt looks like we won't take any action on this issue.
The thinking is that technically Windows NUL: is a file that isn't ever written to. Using it as an output makes sense, because it's just routing output to a file that will never contain anything.
However, when reading from it using the PATH reader, FME recognizes this as a file and produces the error "The source dataset must be a folder not a file."
Since NUL: is technically a file this is correct behaviour.
If you just want to skip this, you can set the "Ignore Failed Readers" option in "Workspace Parameters" and it should be logged as a problem, but FME will carry on with the translation.
Hope this helps to clarify the situation.
Hi @Mark2AtSafe, Thanks for the response. I am not sure whether NUL: is a file. The semicolon : tells me it is a device, like c:\\ on the computer, but, hey, those were the old days back when we were using DOS and CP/M. And some scripting uses the phrase "type nul: > file.txt" to make sure an (empty) file is created, so it's meant to be readable as well (however never yielding any data except EOF, which indeed makes it a file in that case). However, if the Ignore Failed Readers option is preventing the crash I'll use that.