[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Emmet Hikory]

Update of bug #21115 (project freeciv):

  Depends on: => bugs #21999


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Emmet Hikory]

Follow-up Comment #18, bug #21115 (project freeciv):

And I've completed the review of my rebase, with the following results:

(1) was mitigated with patch #4451, but has other issues.  Bug #21991 raised
for that, but not a dependency of this bug, as the issue is more general than
just present==FALSE.

(2) is ideally a no-op given empty nreqs.  Included in the wider patch #4679
(which is mostly irrelevant to present==FALSE, except for one comment change,
and the underlying motivation)

During preparation of patch #4679, found another case of dependency on
"nreqs", rather than wider requirement processing, raised as bug #21992 (which
this depends upon).

(3) patch #4451 addressed this in a better way (although not textually near my
earlier changes)

(4) root issue raised as bug #21982 (dependency of this bug)

I did not find anything else in my prior notes that appears to remain
unaddressed by the current codebase.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Emmet Hikory]

Update of bug #21115 (project freeciv):

  Depends on: => bugs #21992


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Emmet Hikory]

Update of bug #21115 (project freeciv):

  Depends on: => bugs #21982


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Emmet Hikory]

Follow-up Comment #17, bug #21115 (project freeciv):

I rebased an old WIP branch for this cleanup this weekend, and it seems that
vast majority of the issues I had previously identified have been addressed. 
The only remaining parts in my branch were:

1) aicity:affected_unit_class(), which needs to return a list, rather than a
unit_class (with matching adjustments for all callers), in order to support
requirements sets that match multiple unit classes.

2) removing is_effect_disabled from the public effects API, and removing the
section using it from aicity.c

3) aicity:adjust_improvement_wants_by_effects(), which needs to consider that
the improvement under consideration might be present==FALSE

4) effects:is_effect_useful(), which needs to consider that VUT_IMPROVEMENT
may have present==FALSE

That said, the notes on my branch indicated there were also problems in
helptext (most of which appear to be addressed by bug #21454, patch #4400 or
elsewhere).  I can't remember if I completed an exhaustive listing of the
issues in the notes for this patch, so the above may not be complete.  I don't
quite trust my rebase, and know I hadn't subjected the prior branch to
significant testing, so will want to reinvestigate these changes before filing
bugs for them.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Follow-up Comment #16, bug #21115 (project freeciv):

Recent discussion in patch #4401 highlights that there are still known
outstanding issues in the code, not yet raised as separate tickets.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Update of bug #21115 (project freeciv):

 Planned Release: => 2.6.0  


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Update of bug #21115 (project freeciv):

  Depends on: => patch #4451


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Follow-up Comment #15, bug #21115 (project freeciv):

> Just to make sure we don't do duplicate work: have you done 
> any work to convert our rulesets?
No.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Marko Lindqvist]

Follow-up Comment #14, bug #21115 (project freeciv):

> Hm, I'm not wild about potentially disallowing rulesets on
> stable branches.

True, that would be quite blatantly against datafile format freeze.
log_error() (limited to one client popup even if there's multiple negated
reqs) sounds sensible - in most cases it just makes ruleset author to fix the
ruleset (should be clearly instructed in the message)

> (I suppose autogames with rulesets defined each way would be
> one way to get a clue...)

Just to make sure we don't do duplicate work: have you done any work to
convert our rulesets? If not, I'll create patch for that (not to be committed
yet, obviously, but to be used in testing).

> There's a weak argument which says that since the syntax on
> 2.5 and 2.6 will be different (negated=TRUE vs present=FALSE)

It's quite perfect counter-argument to the main argument for using negated =
TRUE as default; that it will save ruleset authors from updating requirements
when 2.6 comes out.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Follow-up Comment #13, bug #21115 (project freeciv):

> S2_4: nreqs must be used, patch ruleset sanity checking to 
> disallow negated = TRUE reqs 
> Oh, if we are still going to release 2.3.5, sanity check patch 
> planned for S2_4 should go to S2_3 too.
Hm, I'm not wild about potentially disallowing rulesets on stable branches. An
ultra-stable 2.3 update is not much use if the ruleset you were previously
happy with (didn't tickle any bugs) now doesn't load at all.
I'd go for emitting log_error() on these branches, at most -- that allows us
to communicate that it's not recommended, and shows up fairly obviously in the
client. (Similar to what we do when savegame loading goes a bit wonky.)
On 2.5 and 2.6 we still have time to set a hard policy IMO...

> What about S2_5? Is it transition version where both are 
> supposed to work?
That really depends on whether we can make present=FALSE reliable in time.
Right now I don't think we've scoped out how much work there is to get there
from here -- all the stuff I've fixed, I happened to spot on the way somewhere
else, I haven't done an exhaustive survey.
(I suppose autogames with rulesets defined each way would be one way to get a
clue...)

> maybe it's simply too late in stabilization to make 
> present = FALSE the default now?)
There's a weak argument which says that since the syntax on 2.5 and 2.6 will
be different (negated=TRUE vs present=FALSE), we should only start encouraging
this style once the syntax has settled (i.e. in 2.6) to avoid two lots of
ruleset rewriting for third parties.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Marko Lindqvist]

Follow-up Comment #12, bug #21115 (project freeciv):

Oh, if we are still going to release 2.3.5, sanity check patch planned for
S2_4 should go to S2_3 too.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Marko Lindqvist]

Follow-up Comment #11, bug #21115 (project freeciv):

For effect reqs & nreqs, I think the plan now is:

S2_4: nreqs must be used, patch ruleset sanity checking to disallow negated =
TRUE reqs
2.6: nreqs are to be removed (or only deprecated?). present = FALSE reqs must
be used.

What about S2_5? Is it transition version where both are supposed to work? Or
do we handle it like S2_4? If both are supported, which method is the
preferred one (especially for new rulesets, should we recommend
safe-and-reliable nreqs, or less-work-in-future-update present = FALSE. Also,
supplied rulesets should use what ever is the preferred method - maybe it's
simply too late in stabilization to make present = FALSE the default now?)

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Update of bug #21115 (project freeciv):

  Depends on: => bugs #21430


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Update of bug #21115 (project freeciv):

  Depends on: => bugs #21432


___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev

[Freeciv-Dev] [bug #21115] [metaticket] Negated requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

[Jacob Nevins]

Update of bug #21115 (project freeciv):

 Summary: [metaticket] Negated requirements for effects
('negated'=TRUE, 'present'=FALSE) do not work => [metaticket] Negated
requirements ('negated'=TRUE, 'present'=FALSE) do not work reliably

___

Follow-up Comment #10:

Expanding this metaticket to be about all bugs in negated requirements,
because that's what I've already hooked up, on the understanding that bugs in
non-effects negated requirements are more severe. We can split it again later
if necessary.

___

Reply to this item at:

  

___
  Message sent via/by Gna!
  http://gna.org/


___
Freeciv-dev mailing list
Freeciv-dev@gna.org
https://mail.gna.org/listinfo/freeciv-dev