Ticket #163 (Fixed)Sat Feb 23 20:15:51 UTC 2008

Ugh. That formatting’s hideously broken. Someone tell me how to make pre-formatted text display correctly, and I’ll try again. I can’t find any documentation anywhere about what markup is permitted in bug reports.

Let’s try pre:
<pre>
Here’s some
preformatted
text.
</pre>

We knew about Free – there are other issues – but we didn’t have time to solve it. I also noted that Japan doesn’t build but other territories were fine so again, I was happy to go with that. It’s worth having these things noted in a bug report though, to keep track of fixes later, so many thanks for the report.

The Japan/UK/USA directories are meant to be empty, sort of; it’s a CVS thing (you can delete files but not directories – easily, anyway – so moved components tend to leave their empty trees behind). Hourglass2 needs populating, because it previously only had NC-specific hourglass resources, along with a copy of Doggysoft’s !HourMake modified to produce 32-bit compatible code. We need permission to distribute that.

As for the others – the formatting doesn’t help, but I don’t think we’ve noticed those specifically. Comment formats use Textile, like in the Wiki, forum or news comment sections. The test you did with ‘pre’ ought to have worked, but it’s possible you needed to put the item on a separate line:

<pre><code>This is a test using ‘pre’ with ‘code’ nested inside it.</code></pre>

If not, perhaps the comment section in the tracker filters too aggressively – you might consider filing a bug report for that if it bothers you! – perhaps for longer commentary you could consider starting a forum thread and placing a link to that thread in the tracker comment instead.

For further reference: http://hobix.com/textile/

A really useful preview tool: http://www.textism.com/tools/textile/index.php

Ok, let’s see if textile will build us a table, as that’s probably more use :)

Somewhat better, I guess — borders would help, I suspect.

Anyway, thanks for the reply.

The code that causes issues in the Japanese territory doesn’t appear to actually be called from anywhere. Based on that, I’d just make it use EXIT rather than EXITS (i.e. don’t preserve flags).

CVS is a pain with directories, so that’s understandable. Perhaps it’d be worthwhile to remove these from the list of components released in batch 3? (http://riscosopen.co.uk/content/downloads/risc-…)

I think it’d be worth linking both that guide and the preview tool from somewhere prominent (I’d suggest the bug report form but, as textile also applies to the forum, I’m not convinced that’s the best place). I’ll create another ticket for that so it doesn’t get lost amongst unrelated batch 3 build issues :)

It’s true that the UK, USA and Japan territories were merged into one source tree and consequently removed from the trunk – hence the empty directories (CVS doesn’t prune empty directories from the trunk by default and in some cases components are written on the assumption that pruning isn’t done and will fail to build if it is, so we don’t explicitly prune them either).

The USB issue – the USB sources are unique amongst the RISC OS sources in that the directory from which you start building the component (HWSupport.USB.NetBSD.build) isn’t the root of the source hierarchy. This appears to have caused confusion during the creation of the sources tarfile – the missing directory is HWSupport.USB.NetBSD.dev, as you’ll see from the online CVS repository. In fact, there are a whole lot of other USB directories that were missed out of the tarfile; we’ll have to get that fixed. In the meantime, you should be able to do

The “EXITS without EntryS” warning suggests that the Japan-specific code hasn’t been 32-bitted, so it’s worth double-checking it for any other issues (TEQP instructions, assuming SWI calls preserve flags etc).

The rest appears to be down to deficiencies in the BBE – either missing files or overaggressive pruning of private allocation information (probably down to allocations being made to Tematic or Castle when they needed to say Acorn in order to prevent breaking the BBE). I expect we’ll need to issue an updated BBE.

We could do with checking this now that we’re up to Batch 4 – it’s likely that some of these issues have been fixed.

and yet more checking in Batch Five.

Are any of these still outstanding or can this be closed?

All these components are involved in daily builds so are implicitly fixed.
I think I fixed Japan territory to assemble.
Hourglass2 now contains sources.