Page 1 of 2
Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sat Dec 09, 2023 2:11 am
by ql_freak
It's on
Dilwyns page.
Search for "qvfs06.zip".
I have never tried it albeit I know the author. But if it works, it would be the
MOST IMPORTANT feature, which should be implemented into QDOS (SMS and how they are all named).
It's most probably not possible to port a C++ Compiler, because even for standard headers and libraries the pathname length from QDOS (36 chars [device name is not counted :‑)] is too low, to compile a C++ Compiler (without a lot of change – because of too short pathnames – in the source code).
It's a real problem (36 character length), because this limit is in the access layer of the directory device driver of QDOS. So it's not easy to overcome it.
BUT I THINK: We must find a solution! (Even Microsoft found a solution for FAT file system to overcome the 8+3 filenames).
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sat Dec 09, 2023 11:23 am
by Martin_Head
I wrote this
https://www.theqlforum.com/viewtopic.php ... ept#p27694 some time ago.
It was just a proof of concept, and I never went any further with it.
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sat Dec 09, 2023 3:15 pm
by ql_freak
Thank you, looks interesting. But it's a work of your own right? Not "long filenames" from Recktenwald, which claims to support pathname length up to 255 characters.
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sun Dec 10, 2023 10:40 pm
by Peter
ql_freak wrote: Sat Dec 09, 2023 2:11 am
It's most probably not possible to port a C++ Compiler, because even for standard headers and libraries the pathname length from QDOS (36 chars [device name is not counted :‑)] is too low, to compile a C++ Compiler
A compiler can be cross-compiled on a system without that limitation. I think there are other, bigger problems when it comes to porting modern toolchains.
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Mon Dec 11, 2023 9:34 pm
by ql_freak
Of course you're right. But there are a lot of other programs, which will not run on a QL just because of the (now!) too short pathname length. The problem is of course, that this limit is not in the physical layer of directory device drivers, but in the access layer. In fact I think this is a real design flaw of QDOS, cause you cannot (at least not easy) correct it with a new directory device driver, but only by extending QDOS with a new (second) access layer open routine. And even then, this would only work for (new/ported) programs.
An example would be Python, which would be a very nice add-on to QDOS. And as Python is an interpreter like SB, you can compile it on another OS for QDOS, but it will be (most probably) not fully usable on QDOS, cause there may be import statements necessary which will exceed the QDOS pathname length.
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sat Feb 10, 2024 4:57 pm
by Derek_Stewart
Hi,
The problem with QDOS is restricted to 32 character file names.
With the introduction of Level 2 file system, which introduced the concept of a Hard Directory, which has a file type of 255.
I am wondering if the Level 2 directories could be updated to hold 32 character file names for each directory and subdirectory.
Now, I have asked this before and always been this was not possible.
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sun Feb 11, 2024 4:02 pm
by NormanDunbar
Hi Derek,
I remember many years ago -- which is unusual for me, remembering stuff -- having a rant about this very matter. Why does the QL's file system
need to include the whole directory tree in each and every filename in the directory?
The eRant is at
https://normandunbar.github.io/RantsAnd ... le-system/ and dates from May 2009!
Mind you, if we did do it properly (!) then I suppose we'd be getting into Linux territory and have to use inodes then.
Cheers,
Norm.
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sun Feb 11, 2024 5:02 pm
by tofro
NormanDunbar wrote: Sun Feb 11, 2024 4:02 pm
I remember many years ago -- which is unusual for me, remembering stuff -- having a rant about this very matter. Why does the QL's file system
need to include the whole directory tree in each and every filename in the directory?
The answer to that is simple: Because directories on the QL are just an illusion (no joke intended).
Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sun Feb 11, 2024 5:14 pm
by NormanDunbar
I know!

Re: Has anyone ever tried "long filenames" from H.P.Recktenwald?
Posted: Sun Feb 11, 2024 8:05 pm
by Derek_Stewart
Not much of rant...
NormanDunbar wrote: Sun Feb 11, 2024 4:02 pm
Mind you, if we did do it properly (!) then I suppose we'd be getting into Linux territory and have to use inodes then.

I do not see the problem with this.
But the Amiga has 31 character filename per directory.
Maybe using using a modified file name substitution for the directory tree to a device, like SUB implements. But this would need a change to the operating system, DV4 ?
tofro wrote: Sun Feb 11, 2024 5:02 pm
NormanDunbar wrote: Sun Feb 11, 2024 4:02 pm
I remember many years ago -- which is unusual for me, remembering stuff -- having a rant about this very matter. Why does the QL's file system
need to include the whole directory tree in each and every filename in the directory?
The answer to that is simple: Because directories on the QL are just an illusion (no joke intended).
All of life is an illusion, due to the lack beer...