On Fri, 14 Aug 2009 23:10:53 -0000, wibbled on for an age:
> Author: tlsa
> Date: Fri Aug 14 18:10:53 2009
> New Revision: 9306
> URL: http://source.netsurf-browser.org?rev=9306&view=rev
> Merge from Paul Blokus' selectscroll branch. Fixes text input scrollbar behaviour.
Unfortunately that doesn't appear to have worked. I'm still getting
the same crash in the same place when typing over line ends in text
I obviously didn't check this in the branch.
Typing a line of text in a text field, when it goes on to the next
line (and potentially adds a scrollbar) the attached crash occurs in
Occurs at desktop/scroll.c:437
I've been umming and arring about responding to this for a few days, but basically I'd like to offer some help if I can. I noticed that Steve Fryatt also said he was interested, and he has a _lot_ more experience to bring to the team than me, but I might in time be able to bring something useful as well. I don't have any published software - I've always spent too long planning things and not enough actually doing them!
I have to warn you all:
1) I haven't been a serious programmer since I worked for Oregan Networks about 6 years ago. Albeit that was on their browser (what RISC OS users know as Oregano 2) and was in C, so very relevant to Netsurf. However, I was involved in porting (to the PS2 - but mainly back-end stuff, very little GUI) and core functionality like encoding support, so that's not so relevant to what's wanted here. I do have some experience in WIMP programming, but that was even longer ago! Nevertheless, I've tried to keep relatively up-to-date with the developments in the RISC OS WIMP as it's moved forward.
2) I have no experience of GCC or UnixLib...yet!
3) I have very little spare time. My main job is translating, and, as I work from home, it tends to eat up all of my time! So I doubt I'd be able to put in more than a few hours a week on Netsurf. The first few months are almost certainly just going to be me trying to figure out in some way how Netsurf works, with no useful input at all!
4) I only have an a9home and VRPC to test on. No Iyonix and no RISC PC (I have one, but I won't be able to get it for about 6 months I should think).
Nevertheless, I'm willing to try! I will have a stab at getting a build environment set up next week as a first milestone (I see from recent discussions that building on RISC OS has been made significantly easier just lately).
And that's that. I'd be really disappointed to see Netsurf disappear on RISC OS. It's a great piece of software.
Will Blatchley (wpb)
On Thu, 13 Aug 2009 18:40:09 -0000, wibbled on for an age:
> it looks as though the case of an incomplete fetch is an error
> - FETCH_ERROR_NO_ERROR);
> + FETCH_ERROR_UNKNOWN_CURL_ERROR);
No, that's not right.
An incomplete fetch actually signals the end of the file. The
requested amount (1K in this case) is always returned, unless there
aren't that many bytes left in the file.
FETCH_ERROR_NO_ERROR is therefore correct.
Although technically if FRead returns -1 we do have an error, so that
code could be improved.
Since the last progress report I have completed the hotlist and added
the sslcert window. I also removed all known bugs as well as
unintuitive tree behaviour. There are two remaining things to be done
before merging - making the branch work for riscos and reviewing the
code :). I am not sure what the preferred order here is.
The next things I would like to work at are the select and scroll
widgets. I decided to work on them in one branch because they are
pretty much tied together - the select widget needs to be scrolled and
the scroll has to be tested somehow.
The NetSurf Developers are looking for new developers to join the team.
Our current development plan is outlined loosely here:
We'd welcome any development help but there are two specific areas in
which we need help.
The RISC OS front end
The RISC OS front end was the original front end and in many ways it is
still the most complete. However, the RISC OS front end has not had any
maintainer for over two and a half years. This is now seriously impeding
development of NetSurf's multi-platform core, because the front ends need
to be updated to use core features, and prompting developers to consider
dropping RISC OS support.
Many of the features that were in the RISC OS front end are now being
moved to the core, so that there will be less RISC OS specific code, and
that code can be tested properly on platforms with better debug tools.
If you are interested in working on NetSurf's RISC OS front end, please
get in touch with the developers. We're keen to help anyone get to grips
with the code. Knowledge of C and some experience with the RISC OS WIMP
The NetSurf autobuilder currently builds development builds of the RISC OS
front end every 15 minutes, if there has been a change to the code. It
then uploads the new builds to the web site. It is all done by a Perl
The autobuilder could be made a lot more useful for developers. It could
run the test suites for all components, such as Hubbub and update a page
that shows whether a build has failed or an update has caused a test to be
failed. It could also build NetSurf for various platforms. This would let
developers see if their changes have broken something. For example a Linux
developer may make a change which stops the Amiga version from compiling.
There are some more specific requirements for the autobuilder here:
If you're interested and have some knowledge of scripting languages
(specifically Perl), please get in touch.
If you would like to help in either of these areas, please get in touch
to discuss things with us. The NetSurf IRC channel is best, or use the
netsurf-dev developer mailing list. Details on:
The NetSurf Developers
Michael Drake (tlsa) http://www.netsurf-browser.org/
I've had coffee, checked I'm looking in and compiling the correct
directory, and running the executable I should be, and have come up
with a solution (attached).
I'm not entirely happy with it. It looks like the core is using
current_redraw_browser somewhere outside of content_redraw() -
previously it was pretty happy with being NULLed after
content_redraw() had returned.
The second issue is that with knockout plotters enabled,
knockout_plotters_start() gets called a second time, after plot has
been reset to amiplot (it may be a bad
knockout_plotters_start/knockout_plotters_end pairing). This asserts,
so I've set knockout to false for now.
I suspect both of these issues are linked, but I don't know enough
about the core to figure out exactly what is going on.
The firm "pencils down" date (Monday, 17th) for this year's GSoC is
almost upon us. In preparation for this, I'll try to summarise where I
think the relevant projects are. Please correct any inaccuracies.
1. Bo Yang -- libdom
DOM Core -- branch merged to trunk.
DOM Events -- branch in preparation for review and merge.
DOM HTML -- initial planning started.
2. Mark Benjamin -- GTK improvements/Win32 port
GTK improvements -- branch approaching readiness for merge.
Win32 port -- basic functionality implemented.
3. Paweł Blokus -- core widgets
Textarea widget -- merged to trunk.
Treeview widget/Cookies/Hotlist/Global history -- blocked on RISC OS.
Select widget/scrollbars -- had initial review.
I fully expect to see the DOM Events, GTK improvements, and Select
widget/Scrollbars branches merged into trunk before the end of this
The Treeview widget and associated UI will be merged as soon as the RISC
OS frontend can be fixed up to work with it.
The Win32 port needs an initial review.
I'd like to take this opportunity to thank Bo, Mark, and Paweł for their
hard work this summer and hope you'll continue to contribute to NetSurf
after GSoC is finished.
Finally, I will be away from the 14th until the 18th. I can make no
guarantees as to my connectivity between those dates. Therefore, if you
need me to review anything with the intention of getting it merged into
trunk before the 17th, then it needs to be ready within the next couple