Some feedback

[Waldir Pimenta]

On Tue, Jun 25, 2013 at 8:25 AM, Joachim Breitner
<mail at joachim-breitner.de>wrote:

> Hi,
>
> Am Dienstag, den 25.06.2013, 02:47 +0100 schrieb Waldir Pimenta:
> > That's excellent, thanks a lot! I do agree that arbtt-dump should be
> > kept separate from the categorization, but on the other hand I assume
> > it will be faster for arbtt to output a subset of its entries directly
> > than to, say, pipe the whole output to grep. Could there be a simple
> > grep-like (regex support would be ideal but simple string matching
> > would already be great) ability to limit arbtt-dump's output to
> > entries whose program or title match a given search?
>
> I see the need. Adding reges support or anything the like will just lead
> to something similar than what arbtt-stats does. I guess I could
> refactor the code that the non-tagging part of the categorization
> language is available to arbtt-dump. But from there it is but a small
> step to allow the language to define tags, and then people want to use
> the rules they have in categorize.cfg, and then arbtt-dump turns into
> arbtt-stats.
>
> The solution could go the other way: Add a --dump option to arbtt-stats
> that acts like another report and outputs the selected samples.
>

Hmmm... I'm no sure that would be more intuitive. Wouldn't that end up
doing the reverse (making arbtt-stats behave like arbtt-dump)? Mind you,
I'm not sure that's necessarily a bad thing, just saying. In the end, I
guess both utilities converge; that is, they might be seen as representing
two extremes of a spectrum between a full, raw dump and a fully processed
output; something like this: arbtt-dump --> arbtt-dump --last -->
arbtt-dump --format --> arbtt-dump --filter (or arbtt-stats --dump) -->
arbtt-stats. Does that make any sense? And if so, do you have any ideas
about what should be the overarching strategy moving forward?

>
> Oh, an idea for the homepage: A section testimonials, where users (like
> you ;-)) describe in one short paragraph what they use arbtt for and if
> they have found something interesting.
>

I see what you mean, but "testimonials" sounds a little "corporate" to me
(like we're trying to sell something!). Do you think the quickstart section
of the "usage" tab could serve for that purpose, by including more common
usage patterns? For example, I recall seeing in the mailing list someone
asking for time-based filtering, which, IIRC, was later integrated into
arbtt-stats.

I do, however, think we could include a little about arbtt's history,
perhaps as an introductory paragraph in the "get involved" page, to allow
people to understand who's behind the project. I was thinking something
like "arbtt was created in 2009 by Joachim Breitner (nomeata) <link to the
blog post that introduced arbtt> and as it developed over the years, it
also received contributions by several of its users (here, perhaps link to
the contributors section of the docs)". What do you think?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.nomeata.de/pipermail/arbtt/attachments/20130625/abb9c951/attachment.htm>