Forum software on !WebJames
Pages: 1 2
Rebecca (1663) 107 posts |
Is it possible to run some sort of Forum software using !WebJames? It doesn’t need to be fancy and would only be used on the local network. Cheers. |
Rik Griffin (98) 264 posts |
There are several forum systems out there written in PHP, but I suspect they all need a back end database to store the posts etc in. Unless someone has ported, say, SQLite to RISCOS you might be out of luck. It’d be an interesting project though, I wish I had time to look in to it more. |
Wouter Rademaker (458) 197 posts |
This looks like a list of php flat file based forum software: |
Steve Pampling (1551) 7962 posts |
People should note that the webjames php handler (and the perl handler) are not ARM7 safe. It may work on the Pi with appropriate settings (I haven’t tested) but the beagle fails – loose diagnosis is that the UnixLib element was using dodgy settings and needs recompiling. I started looking as an exercise in learning to do such things and then the two man network team here at work became one. 24/7 on-call anyone? |
rob andrews (112) 200 posts |
will this do what you want http://www.reallysmall.co.uk/Pages/normal/software/database/odbc/odbc.html |
Rebecca (1663) 107 posts |
Thanks for the info guys. I’ll have a play at the weekend. Cheers. |
Chris Gransden (337) 1148 posts |
You can download an armv7 safe version of the WebJames php handler here. I also updated it to the most recent version of php 5.2. From 5.2.2 to 5.2.17. As it’s built with a more recent version of gcc it should hopefully be a bit faster.
SQLite support was already built into the php handler. You can see what is available by browsing to localhost://info.php. |
Steve Pampling (1551) 7962 posts |
Something to play with at the weekend – assuming I don’t end up doing things in the garden. |
John Sandgrounder (1650) 574 posts |
I have been trying to transfer a 126K data file from a Raspberry Pi running RISCOS (RC8) to a Raspberry Pi running Raspbian Wheezy. The best results I have have had use WebJames as the server (on RC8) and wget as the client (on Raspbian). But it still fails (with a corrupted file) about 30% of the time. I saw the post about the Arm7 safe version of Webjames and thought – Just what I need. But no luck – it still has a 30% corruption rate. Has anybody got any ideas, as this is driving me crazy :( |
Wouter Rademaker (458) 197 posts |
Try NFS, Sunfish and/or Moonfish on RISC OS, or FTP, FTPc is a good RISC OS FTP client. and read https://www.riscosopen.org/wiki/documentation/show/Beginners%20FAQ:Network,%20Internet%20and%20Email |
Dave Higton (1515) 3409 posts |
Does anyone have any idea why file corruption is happening at all? |
neil.r (1738) 66 posts |
There’s also accesspluspy which I use regularly to transfer from RISC OS to Raspbian: https://bitbucket.org/mjwoodcock/accesspluspython-dev – works well for me, but needs a kick to restart it every few days. |
Rick Murray (539) 13432 posts |
That is what I was wondering. You were using http on top of TCP – so in theory….. ;-) |
John Sandgrounder (1650) 574 posts |
To Wouter Rademaker and Neil.r, Many thanks for your suggestions. Transfering a file manually between any of my computers is not a problem. I have many options, including as you suggest, NFS, Sunfish, Moonfish, ShareFS and others. But what I need is the two computers to run un-attended for days at a time and for the transfers not to lock up either computer in the case of any fault. Unfortunately file sharing on a RISCOS computer is not that robust, because of the co-operative multi-tasking operating system. (we all see the ‘Please Insert Disc’ error message which stops the system). It is also a requirement for my multi-computer project that the RISCOS computer is the server on which the file is created and the Raspbian Linux is the client where the file is to be analysed. Using an FTP server (FTPs) on the RISCOS Pi computer works as (un-)reliably as WebJames. To Dave Higton and Rick Murray, I think that it is not likely to be just me that would benefit from having a solution to this problem. As you say, it should work but it does not (either with HTTP or FTP). My software has all been running without problem on RISCOS and Linux for the best part of twenty years and it still works on all combinations of hardware which I have tried; except RISCOS on a Raspberry Pi talking to Linux on a Raspberry Pi. I can perm any compination of systems from Real Acorn A410, A5000, RiscPC and Virtual Acorn with numerous Linux systems on a variety of hardware including Sun Sparc. They all work. It even works with RiscOS on Raspberry pi talking to any other Linux – or on Real or Virtual Acorn talking to Raspbian on a Raspberry Pi. But it does not work reliably on a Raspberry Pi (RISCOS RC6 or RC8) talking to Raspberry Pi (Rasbian), which is what I would like to use, because of the low power requirements and small footprints. It may be my imagination but I think that RC8 gives a small improvement over RC6. Can anyone else repeat the fault? – The webserver file can be data or HTML – both suffer corruption if the file big enough. |
Rick Murray (539) 13432 posts |
Well, that is a head scratcher. Anything to Raspian works, anything to Pi-RISC OS works. Pi-RISCOS to Raspian fails. How utterly bizarre! If you have some time (and inclination), it might be interesting to try Ri-RISCOS to Pi-RISCOS and Raspian to Raspian; just in case there is some sort of ‘quirk’ when communicating between two Pies. It’s a long shot, but I don’t otherwise see any logic behind the two halves working while together they fail. Sorry, I can’t attempt to replicate this. I don’t have two Pies.
…isn’t that usually down to:
|
Dave Higton (1515) 3409 posts |
The only way I can think to do the investigation is to start with a Wireshark capture of the traffic and compare with the file contents. Yes, I know that’s laborious. One of the things I do at work when investigating is to write my own decoder for a pcap or pcapng file, tailored to the purpose. It’s not that difficult; I’ve done one in BBC BASIC. But in your case it may be possible to use Wireshark’s own facilities to extract the data content from the transfer. There are a couple of free file comparison applications available for RISC OS. It would help to know what form the file corruption takes; are bytes missing, are bytes repeated, are bytes inserted, or are bytes changed? How many? |
neil.r (1738) 66 posts |
hmmm, this is all a bit curious, as it’s pretty much exactly what I do, regularly, with no corruption at all. As an example, I use !SafeStore to backup my RISC OS data every night to an external HDD mounted on my Raspbian machine, with no problems at all. As I mentioned earlier, every now and then I need to restart the access.py process as it’s fallen over somewhere along the line, and I’ve never investigated that as I’m not a python programmer. Come to think of it, I’m not a programmer at all. But, I’m willing to bet if I put together a cron job that automatically restarted the access.py job every day, the problem would go away. I’ve also mounted the same drive using !OmniClient when it’s exported by the Raspbian server as a Samba share. Again, no problems at all with any kind of file corruption. I’ve also used FTP (although with the Raspbian machine acting as server) again, no problems at all. If there’s a specific test I can help you with, just let me know – for the record, my Raspbian version is 3.6.11+ #371 |
Martin Bazley (331) 379 posts |
John: I’m clutching at straws a bit here, but just on the off-chance, are you using an encrypted (sftp or https style) connection? Furthermore, are any of the programs you’re using calling R-Comp’s SecureSockets module? I’ve found that that module is very prone to corrupting files during transmission (oddly, receiving seems fine). It depends on the protocol used – I had very different results from two different SMTP servers, but both were internally consistent – but as a rule, it’ll successfully transmit the first N bytes (where N varies; it was consistently around 100KB for one server) but then accidentally miss out a block of 1000 or so bytes. This pattern repeats cyclically if the file is longer than 2N bytes in total. I was testing with sending emails rather than transferring data, which introduced interesting complications. For example, the effects of base64 encoding meant that when the file was decoded at the other end, sometimes all the remaining bytes after a ‘jump’ would be 4 bits out of alignment! What exactly do you mean when you say “corruption”, anyway? Might be worth comparing your symptoms to mine, although I don’t know the internal details of the protocols you’re using, so further advice on what to look for may not be very useful. |
John Sandgrounder (1650) 574 posts |
Hi Martin: The software configuration I refered to above used a standard RISCOS RC6 (or RC8) build on the Raspberry Pi with just !WebJames (server) added. Whilst I have been quiet on this forum, I have run another four FTP tests with results as follows: !Deltanet FTP Server on Virtual Acorn serving to FTP command line client on Solaris – Works OK In all four cases, I used the same 126K file transfered in Binary mode. An example of the “corruption” is given below (all lines of this test file should be the same) 01234567890——abcdefghijklmnopqrstuvwxyz——ABCDEFGHIJKLMNOPQRSTUVWXYZ and 01234567890——abcdefghijklmnopqrstuvwxyz——ABCDEFGHIJKLMNOPQRSTUVWXYZ Hi Neil.r I am not trying to transfer the file with any mounted or shared drive. But if you have the time, I would like to take you up on your offer of help. Can you try the one or both of the following tests for me, with a file of about 126K bytes. !WebJames server running on Raspberry Pi RISCOS My Raspbian is version 5.8.11 (But I have also tested this with a later one) |
neil.r (1738) 66 posts |
So far, so good… Using !FTPs v0.24 on RISC OS, and a command line FTP client from Raspbian, have transferred a 129K image file six times in a row, with no problem at all. Not yet tried !WebJames, but I’ll give it a whizz a bit later. |
John Sandgrounder (1650) 574 posts |
Hi Neil, Many thanks for that. I should have added that in my tests, the files appear to transfer OK, but contained errors. I transfered two identical files (checked with diff) and then checked the results (also with diff). Some transfers were OK but on most the diff failed. Please accept my apologies if you have already checked the file contents by any means. I am out until Sunday, but I will repeat my tests with Raspbian version 3.6.11+ #371 (or later). Which RISCOS build are you using? |
neil.r (1738) 66 posts |
Ah, yes, I use diff to check the files as well :) I’ll keep trying, as like I mentioned, I’ve only done a few transfers so far, so it’s not by any means comprehensive. I’m using the latest version of RISC OS that PackMan grabbed for me from the testing respository. info shows 5.19 (19-Mar-13) if that helps. Ta, |
Dave Higton (1515) 3409 posts |
Looking at one of the examples of corruption, I see “200 Command okay.” Where did that come from? The only way you’ll resolve which end it is, is to run a Wireshark trace. That will tell you whether the corruption was sent (in which case the problem is at the sending end) or not (in which case the problem is at the receiving end). Wiresalmon on RISC OS is perfectly capable of capturing a pcap file that can be examined in Wireshark. |
John Sandgrounder (1650) 574 posts |
Good Afternoon All, Amazing how a face to face conversation can help. I have just returned from a very useful, interesting and enjoyable few hours at the Wakefield show, where Theo on the RISCOS OPEN stand suggested that I changed my network to exclude the 10 Mbit/sec Hub to which the RISCOS pi was connected). Everything now works perfectly on a 100 Mbits/second switch (using either !WebJames or FTPs) Thank you all for your time and comments. My otherwise useful 16 port hub now has a warning label stuck to it. |
Chris Evans (457) 1614 posts |
10Mbit problem now fixed:-) |
Pages: 1 2