Re: Margins

2010-12-04 Thread John Tytgat
In message <517ffe259ach...@chris-johnson.org.uk>
  cj  wrote:

> Well, the !Boot file has the line
> 
>   Set NetSurf$Dir 
> 
> in it, so booting an obsolete version will definitely change the
> system variable. I thought it was recommended that boot files should
> have that line of the form
> 
>   If  = "" Then Set NetSurf$Dir 
> 
> so the variable is only set if it doesn't exist. Only the !run file
> sets the variable directly.

Indeed.  I've commited a fix for that.

Thanks,
John.
-- 
John Tytgat
j...@netsurf-browser.org



Re: Margins

2010-12-03 Thread cj
In article <92cdf77f51.r...@user.minijem.plus.com>,
   Richard Porter  wrote:
> That's what I would have thought, but I can't see how, when you
> inadvertantly boot an obsolete version, you get bugs that were
> present in that version and when you reboot the correct
> application, without quitting and re-running it, it comes all
> right again. This is easy to demonstrate. Maybe that particular
> bug was data-related.

Well, the !Boot file has the line

  Set NetSurf$Dir 

in it, so booting an obsolete version will definitely change the
system variable. I thought it was recommended that boot files should
have that line of the form

  If  = "" Then Set NetSurf$Dir 

so the variable is only set if it doesn't exist. Only the !run file
sets the variable directly.

It means the source of various resource files will be set to the old
version, e.g. the default CSS styles which may influence the layout
of pages.

-- 
Chris Johnson



Re: Margins

2010-12-03 Thread Richard Porter
On 2 Dec 2010 Rob Kendrick  wrote:

> On Thu, Dec 02, 2010 at 02:32:18PM +, Richard Porter wrote:
>> On 2 Dec 2010 Rob Kendrick  wrote:

> The problem is, as you said yourself, we are not sure what the issue is.

>> You didn't answer my question: is it possible for NS to execute old
>> code whilst still announcing itself to be the current version?

> Code, no.  The revision number is encoded inside the !RunImage.  Data,
> it's possible; I've not looked.

That's what I would have thought, but I can't see how, when you 
inadvertantly boot an obsolete version, you get bugs that were present 
in that version and when you reboot the correct application, without 
quitting and re-running it, it comes all right again. This is easy to 
demonstrate. Maybe that particular bug was data-related.

The good thing is that I've done a change to SiteMatch which should 
stop it picking up obsolete data files if the app$Dir variable is 
accidentally changed. The path is stored on start-up.

-- 
Richard Porterhttp://www.minijem.plus.com/
  mailto:r...@minijem.plus.com
I don't want a user experience; I just want stuff that works.



Re: Margins

2010-12-02 Thread Rob Kendrick
On Thu, Dec 02, 2010 at 02:32:18PM +, Richard Porter wrote:
> On 2 Dec 2010 Rob Kendrick  wrote:
> 
> > On Thu, Dec 02, 2010 at 12:09:52PM +, Richard Porter wrote:
> 
> >> Is it possible to avoid this situation, for example by storing global
> >> variables locally at start-up and not relying on them always pointing
> >> to where they should be pointing to?
> 
> > Alternatively, you could fix your back up software (which is clearly
> > broken if it really is at fault), or not have a four-year-old copy of
> > NetSurf hanging around.
> 
> I think that ducks the issue. 

The problem is, as you said yourself, we are not sure what the issue is.

> You didn't answer my question: is it possible for NS to execute old 
> code whilst still announcing itself to be the current version? 

Code, no.  The revision number is encoded inside the !RunImage.  Data,
it's possible; I've not looked.

B.



Re: Margins

2010-12-02 Thread Richard Porter
On 2 Dec 2010 Rob Kendrick  wrote:

> On Thu, Dec 02, 2010 at 12:09:52PM +, Richard Porter wrote:

>> Is it possible to avoid this situation, for example by storing global
>> variables locally at start-up and not relying on them always pointing
>> to where they should be pointing to?

> Alternatively, you could fix your back up software (which is clearly
> broken if it really is at fault), or not have a four-year-old copy of
> NetSurf hanging around.

I think that ducks the issue. I don't need to have a four year old 
copy of NS hanging around, and it won't be once I've archived it to 
the network drive. In fact I don't need to keep it at all but it was 
there. However I do need to keep other more recent copies of NetSurf, 
for example the last one or two development versions in case of 
problems, and the last released version.

As for the backup software, I am chasing down problems with SafeStore/ 
LanMan98/network drive not coping with filenames that contain 8-bit 
characters and crashing the drive. I don't know why the backup system 
causes !Boot files to be executed but I had one instance of an error 
because a filename in a !Boot file did not use the same case as the 
actual filename, and the network drive filing system is case 
sensitive.

Anyway, given the propensity of RISC OS to boot any old application 
that it comes across, it follows that running applications should be 
immunised against this possibility.

You didn't answer my question: is it possible for NS to execute old 
code whilst still announcing itself to be the current version? I 
assume the answer is 'yes' because running the correct !Boot file 
corrected the fault without having to quit and rerun NS.

-- 
Richard Porterhttp://www.minijem.plus.com/
  mailto:r...@minijem.plus.com
I don't want a user experience.  I just want stuff that works.



Re: Margins

2010-12-02 Thread Rob Kendrick
On Thu, Dec 02, 2010 at 12:09:52PM +, Richard Porter wrote:

> Is it possible to avoid this situation, for example by storing global 
> variables locally at start-up and not relying on them always pointing 
> to where they should be pointing to?

Alternatively, you could fix your back up software (which is clearly
broken if it really is at fault), or not have a four-year-old copy of
NetSurf hanging around.

B.



Re: Margins

2010-12-02 Thread Richard Porter
On 2 Dec 2010 Rob Kendrick  wrote:

> On Wed, Dec 01, 2010 at 11:40:21PM +, Richard Porter wrote:
>> NetSurf has suddenly started to insert blank space around the edge of
>> my pages so that tabs that were nicely against the top edge are now
>> hanging in mid air. r10948 was OK; r10951 is exhibiting the fault.

> Can you either check this again, or provide a minimum test case?
> Because none of the changes between these two revisions touch layout.

It's happened again, and I'm backing up files again. I have a theory 
that might be connected to what occasionally happens in SiteMatch.
The directories I copied included old versions of NetSurf and 
SiteMatch, and it's possible that RISC OS could have booted those 
applications.

Although the info on the icon bar still says "r10951", is it possible 
that NetSurf is actually picking up obsolete code from the backup 
file? For certain NetSurf$Dir now points to a Jan 2006 version of 
NetSurf the backup drive.

Is it possible to avoid this situation, for example by storing global 
variables locally at start-up and not relying on them always pointing 
to where they should be pointing to?

-- 
Richard Porterhttp://www.minijem.plus.com/
  mailto:r...@minijem.plus.com
I don't want a user experience.  I just want stuff that works.



Re: Margins

2010-12-02 Thread Richard Porter
On 2 Dec 2010 Rob Kendrick  wrote:

> On Wed, Dec 01, 2010 at 11:40:21PM +, Richard Porter wrote:
>> NetSurf has suddenly started to insert blank space around the edge of
>> my pages so that tabs that were nicely against the top edge are now
>> hanging in mid air. r10948 was OK; r10951 is exhibiting the fault.

> Can you either check this again, or provide a minimum test case?
> Because none of the changes between these two revisions touch layout.

That's strange because it's not doing it this morning. The site was 
http://www.minimarcos.org . I don't know what might have affected it 
yesterday. The only significant thing I was doing was copying a lot of 
files to a network drive via LanMan98 but that shouldn't have affected 
NetSurf.

Richard
-- 
Richard Porterhttp://www.minijem.plus.com/
  mailto:r...@minijem.plus.com
I don't want a user experience.  I just want stuff that works.



Re: Margins

2010-12-01 Thread Rob Kendrick
On Wed, Dec 01, 2010 at 11:40:21PM +, Richard Porter wrote:
> NetSurf has suddenly started to insert blank space around the edge of 
> my pages so that tabs that were nicely against the top edge are now 
> hanging in mid air. r10948 was OK; r10951 is exhibiting the fault.

Can you either check this again, or provide a minimum test case?
Because none of the changes between these two revisions touch layout.

(They are: improve the function that identifies IP addresses from host
names, an AmigaOS frontend change, a fixed include for our GIF decoding
library, set the focus to the main window widget by default on startup).

B.



Margins

2010-12-01 Thread Richard Porter
NetSurf has suddenly started to insert blank space around the edge of 
my pages so that tabs that were nicely against the top edge are now 
hanging in mid air. r10948 was OK; r10951 is exhibiting the fault.

-- 
Richard Porterhttp://www.minijem.plus.com/
  mailto:r...@minijem.plus.com
I don't want a user experience.  I just want stuff that works.