Text files in source release

Is it normal for the text files in the source release to be typed as ‘data’ or is it the way you are packaging ithe source? It’s a bit of a chore working out if any are supposed to be data before changing them to text.

I suppose what I’m really asking is any chance of having the correct filetypes in future?

They’re not filetyped anything per se, and whatever you used for Tar has decided to allocate filetype ‘Data’ rather than ‘Text’. I’m guessing you used SparkFS. You need to close the archive (Ctrl+Drag it to the icon bar icon and choose “Forget”, else just quit and restart SparkFS), then issue the following commands:

*SparkFSExtension fff
*SparkFSTruncate 240

The first command associates RISC OS filetype &fff with no filename extension, i.e. it is used for any “untyped” file. The second is for safety and makes sure that SparkFS doesn’t truncate any filenames. You will find that some of the files in the archive exceed 10 characters, so if using an old filing system you will need to take steps to work around that (e.g. move things to an image filing system that supports longer filenames, run !LongFiles if you’ve an older machine, etc.).

Further to this – I’ve added a note on the current sources page. I’m also restructuring it, into a more general downloads section. Should be a lot easier to navigate and will give us more room to add documentation in logical places.

Thanks.

You may also like to note that adding SparkFSExtension fff in run doesn’t work. (SparkFSTruncate can be set from the !run file)

Issuing the command after loading SparkFS did the trick.

I’m not going to bet on this, but I think the first line of !SparkFS.CONFIG.Extensions is the default RISC OS filetype to apply to files without an extension. Certainly, in my copy this line reads:

0xfff

which would explain why I don’t jump through any hoops of calling SparkFSExtension. You might want to check that this file contains all of the other common extensions you can think of – a quick look in your MimeMap file might help. (I suppose SparkFS doesn’t use the MimeMap?)

I do, however, have a line saying:

SparkFSTruncate 255

in !SparkFS.!Run which will be helping to avoid problems with long filenames in archive getting truncated (and thus breaking the build).

Thanks. Changing the first line of !SparkFS.CONFIG.Extensions to 0xfff did the trick. Thinking about it though I think it would be better if text file typing was explicit in the source. Someone else may send me something assuming filetypes default to data. Anyway just a thought.

Like you I already had SparkFSTruncate 255 in the !run file

OK, thanks; I’ve incorporated the above into the note on the tarballs page.