?
?
Alexander Klenin hat am 16. M?rz 2011 um 12:14 geschrieben:
Show quoted text 14 lines
2011/3/16 Mattias Gaertner :
>> > Write a package TAChartAggPas, that depends on AggPasLcl and
>> > tachartlazaruspkg and has a unit implementing the aggpas backend for
>> > TAChart.
>> I considered that. But it will be one package per dependency -- so +3
>> packages right now.
>> Is it ok from user interface point of view?
>
> I think yes. The project requires only one dependency instead of two
> (TAChart+AggPas). So this might actually be an advantage.
>
I was rather wondering about how easy it would be for the user
to discover the need for that one dependency.?
Like any other option. Just add hints/comments to the right places. ?
?
Show quoted text 17 lines
>> Is there a notion of "package group" in the IDE?
>> If not, maybe it is a good idea to create one?
> What should that be?
>
> Packages are shown alphabetically, so if you use the same prefix they are
> listed close together.
I was thinking about the following:
1) Let "sub-package" be a special kind of package with the additional
"Parent" field
2) In the IDE package list, display only a parent packages (maybe
display sub-packages
in a treeview-like manner)
3) Automatically install sub-packages if all their dependencies are installed
(that, naturally, includes parent packages).?
The 'install' is only for the IDE and independent of the project. Think about the IDE as another project.
That means 'soft' dependencies can be satisfied by the IDE and/or the project. Which means the package might have different search paths for the IDE and for the project, which might increase confusion.
And you need more IDE knowledge to understand the package.
?
Show quoted text 7 lines
The end result would be, if the user installs TAChart and OpenGL, he will
automatically have opengl back-end, without the need to worry about
TAChartOpenGL package.
But may be it is overkill. To this moment, I am aware of a single user
interested in AggPas back-end.
Yes. To summarize:
- There exists another solution, which is not that bad.
- The proposed solution has some drawbacks.
- So far only one request, which makes maintaining/testing difficult.
My conclusion: a good day work better spent elsewhere.
Mattias
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20110316/608bca7a/attachment.html>