[elinks-users] Thoughts on a stable 0.10 release
Miciah Dashiel Butler Masters
miciah at myrealbox.com
Tue Nov 30 10:48:25 PST 2004
We ELinks developers are all quite busy at our respective universities
or jobs--pasky and I have just started college--and there just isn't
much time between the end of the semester and December 25.
This being so, Jonas Fonseca pronounced on IRC that we would be unable
to make a stable 0.10 release for Christmas.
Being who I am, I had to disagree. (Hm, I wonder whether
he was in fact counting on that...)
N.b.: The following are my opinions and goals. They are not necessarily
those of the maintainer or the project. I'm just one contributor.
I invite others to comment and express their own desires--hence
this public E-mail.
Here are some criteria for a possible release in the little time
we have left:
- there must be no regressions from 0.9;
- we should have up-to-date translations;
- we should update the documentation, at least to avoid
any misinformation and preferably to avoid any decrease
in the available documentation (both could be called regressions);
- we should change _now_ any APIs (options, scripting interfaces,
command-line arguments, &c) that we do not want to keep around
for a while; and
- experimental features must be marked as such and disabled by
default--we would not want a major OS distributor to ship a release
of ELinks with ECMAScript support that opens it up to DoS attacks,
or CSS support that significantly hurts performance (these are
two known problems--see bugs 545 for the problem with CSS
and query for ECMAScript bugs to see the possible DoS attacks)
I constructed a Bugzilla query for bugs that have appeared since the
30 November release of ELinks 0.9, excluding enhancements and changes
to optional new experimental components (i.e., CSS, ECMAScript,
and the SGML engine):
I would like:
- to check for each report whether the bug has been fixed
but the report left open;
- to prioritise bugs that would require string changes to fix
(give the translators a fixed target sooner rather than later); and
- to mark reports of regressions with a special keyword, or per haps
give them 'blocker' severity.
The above is intended to ensure that a new release will not be _worse_
than the current stable. However, there are several tasks that I would
particularly like to accomplish in order to make this release _better_
than the last:
- fix issues found by Michal Zalewski mangle tool--what kind
of people would we be to ship a browser known to be vulernable
to remote DoS attacks?
- improve the performance regression introduced by CSS
(otherwise, very basic CSS works well);
- improve the documentation--how about a goal of documenting
one important feature that currently lacks documentation?
- improve the searching in hierboxes--some hierboxes, such as
the options manager, filter the results rather than jumping
between them, and this is confusing.
I have a few more goals for a release, but I don't think that we can
accomplish these in time:
- fix refresh support (bug 343);
- fix the decompression support (bugs 517 and 534)
- improve the scripting interface (bug 70);
Fixing any these would take too much time or be invasive enough
that they have great potential to destabilise the tree. Feel free
to prove me wrong.
(I shouldn't even bother to mention moving to the SGML engine
or adding graphics support--ain't gonna happen.)
As always, we will need to update NEWS, somebody will need
to write the announcement, and it wouldn't hurt to make sure that
CREDITS is not missing anything.
The release described above might be less exciting than the one that we
envisioned earlier in the year, but I believe that we have a half-decent
chance of completing the enumerated tasks, and I believe that this
would result in a better release than the last, and that is certainly
a good thing.
I would also like to note that even if we cannot get a release out
for Christmas, a New Years release would still be nifty. Hey, I'm
not a big religious nut, I don't get all excited about orbits and
planetary rotation and stuff, I don't really care--but if it means
something to others and can serve as inspiration to get work done,
I'm cool with it.
Miciah Masters <miciah at myrealbox.com> / <mdm0304 at mail.ecu.edu>
More information about the elinks-users