On Tue, Jun 2, 2015 at 4:15 PM, Thierry Goubier <thierry.goubier@gmail.com> wrote:


2015-06-02 15:49 GMT+02:00 Peter Uhn��k <i.uhnak@gmail.com>:
Hum. Alphabetical order on what? That is, to take your example me/my-project, should it be done on 'me' or on 'my-project'?
Hm, that's a good question. It should however be the first letter; so if there is "Uko/QualityAssistant" I would expect it sorted by "Uko"; otherwise there is no much sense in sorting since I can't do binary-ish search.
If it was by the package name, then the name would have to be swapped; "QualityAssistant by Uko" or something like that so I can always look at the first letter.

Ok.
��
��

At the moment, repositories ordered by time of addition in your image has a 'known' behavior: your own repositories come last (and to find them, you just have to look last). Alphabetical ordering would make a mess of that.
Really? Is that something new?
This is a screenshot from Pharo 4.0; but it's possible that I have broken something

The order seems to me at least to be the order in which the repositories were added.

Yes, and I can see that you load many repositories, mixing external software (Roassal) and the things that you seems to be focusing on.

I guess a nice difference here would be to sort there among repositories you have the right to write on and the others... In many (all?) cases, you just want to see the ones you have the ability to save something in (such as when doing a Monticello package save or copy).

That's an interesting idea. In most cases I do indeed care only about my packages, but not always. Sometimes I need to look at other (Roassal) to see what exactly has changed or to track down some new bug. But proper sorting would group my packages together anyway so I'm not sure of the benefits here.
��

��

Could your ordering be solved by just having more readable entries (the entries you suggest)?
I would say that it would improve it regardless of sorting.

Certainly. But I wonder if it wouldn't be enough. A stable ordering allows you to remember where things are (at the beginning, in the middle, in the end, etc...). Alphabetical ordering is not stable.
Why would I need to remember it in the first place? If I am working on more than one product I am not going to remember it anyway.
However alphabetical order _is_ stable. While not absolutely (if I add an extra package it would move), but you will not remember exact index where it is. If I am looking for Roassal then I automatically look at the bottom half (or second third) because I know the relative position of the letters within the alphabet.

Besides with the "fixed" ordering I have to learn every new package where it was again.
��

Example:
- old system browser has a stable ordering, which is alphabetical ordering of base packages, followed by your packages (or the packages you loaded).
- Nautilus has alphabetical ordering of all packages, including yours.

Now, try to answer for a newbie the following question:
- You just created a new package with a name of your choosing. Where it is in the system browser?
�� - Old system browser ... at the end of the first pane on the left.
�� - Nautilus ... euh... somewhere in the first pane on the left.

I for one would actually expect it to be sorted among the others. But that may be just my personal preference. I mean, I see all the packages sorted, why would I ever expect a new package to be suddenly out of order?

This all probably also heavily depends on how much you use it. If you open it rarely you benefit much more from easy-to-understand ordering that I don't need to remember. If I am working with something constantly than the order doesn't matter since I've built a mental map.

But we could make an informal strawpoll to see how (much) people's view vary.


You see? Nautilus has many ways of easily finding that package (using the filter box), but you need to explain them to the newbie before he can use them. With the old system browser, no need to do so.
Well we don't need to treat the newbie like an idiot either; given it's looks and position the functionality is obvious.
But that gives me another idea that would solve my problem - let the Monticello Browser remember my search and give it the same search power (i.e. multiple repositories "repo1|repo2|..") just like Nautilus.

Peter

��

Thierry
��

Peter