On Fri, Jul 23, 2004 at 10:29:37AM +0200, Sebastian Guttenberg wrote:
> Hello everybody!
> I'm working with LyX mainly for writing down long and nasty
> calculations. Very often I calculate directly at the screen, by copying
> the last line and then doing some changes in the copied line.
[Btw M-m w c helps with that...]
> This prevents one from doing sloppy mistakes that one does, if one
> performs the calculations by hand. And this is the reason why I prefer
> LyX to LaTeX. However, for this to work well, it is very important
> that the navigation in the equation (arrays) works quickly and
> intuively.
Indeed.
> Some points have been improved in version 1.3.4 with respect to the
> former version, others unfortunately have become worse! I'll list
> here some points that bother me very much:
>
> 1. if you write indices or powers like a^b and a_b there is put a box
> around it. When navigating through the equation with the cursor, you
> always get stick at these boxes (you need to hit the cursor-key once
> more to get into the box). Navigation thus becomes slower and less
> intuitive.
This depends on your point of view. a^b is a subexpression very much the
same way \frac{a}{b} is. So if one accepts a keystroke to go from 'in
front of the fraction' to 'in the numerator' one possibly could accept a
keystroke to go from 'in from of the power' to 'in the base of the
power'. However, that's not the reason how this 'annoying' behaviour
came into existence.
> I don't see an immediate need for these boxes.
It boils down to limited imagination on my side. 1.2.x mathed pretty
much hardwired not just the navigation behaviour of ^ and _ (it still
like that in 1.3.x), but also provided a highly specialized (read "f'ed
up") data structure to support it. This data structure was part of the
probelems we had with maintainability/extensibility/whatever in 1.2.x
mathed.
> If I really want to put some (I assume they correspond to curley
> brackets in LaTeX),
This assumption is wrong as a look at the exported .tex shows.
You get TeX's curly brackets by typing { in 1.3.x and \{ in 1.4.x IIRC.
> I'd prefer to put them on my own, and not automatially.
You can (and have to if necessary) do that still manually.
> 2. the boxing has another disadvantage: one cannot any longer (as in
> former times) copy index structures. Say I have an object x^a_b and
> want to mark and copy only the indices ^a_b, then always the complete
> box gets marked.
This is indeed no longer possible. But copying the whole 'x^a_b' and
replacing the base is just two keystrokes or so longer and even could be
bound to a keyboard short cut if this is an operation you need
regularly.
> 3. I also find it very annoying that self defined macros pop up when
> you pass them with the cursor. This is again very disturbing for
> navigating.
First of all you can "lock" subexpressions by typing M-i which makes
them act as a unit. So you can go over any macro in a single keystroke
if you want preventing any 'macro pop up'.
> The advantage to the former version is that you always
> recognise a macro, but this could also be done by underlying some
> color or whatever.
Another advantage is that the pop up version enables us to edit nested
macros with multiple arguments properly, i.e. offers an ugly but working
solution in contrast to a nicer but sometimes not working solution.
Of course one could invest some effort to make the common version nicer
while retaining the general version as fall-back. But that's a question
of available resources...
> 4. It is very nice in the new version that the <end>-key does not put
> you any longer at the end of the whole array, but at the end of
> certain structures. However, the combination <ctrl>+<cursor-right> has
> excatly the same function. (Has it?)
I believe so. The code suggests that word-right and end are handles the
same, too.
> I would suggest, to use this latter key-combination to let the
> cursor jump to the next sign, like +,-, and perhaps also \times,
> because sometimes one has long equations with a lot of summands
> without any structures like brackets. I would also find it useful to
> include ordinary brackets ")" instead of "\right)" to the structures
> where the <end> key jumps to.
Good idea. It basically boils down to what is considered a 'word' in
math. 1.3.x asumption is 'a word' == 'a subexpression'. But a term in a
summ or similar makes sense, too. Care to add a feature request on
bugzilla?
> 5. this next point- I think - is a bug: when I'm in an equation array
> within a very large footnote and click in the equation array somewhere
> outside and below the footnote, the screen scrolls down so much that
> the point where I clicked is out of sight! I'd prefer it not to
> scroll at all at this moment.
That's probably a bug but something that's not simply to fix in 1.3.x.
Would be nice if you could check whether this behaviour is still present
in 1.4.0cvs.
> 6. I find it now very nice that undo only undoes the very last change
> and not a complete bunch of changes. However, the screen should if
> possible not scroll when one is doing an undo with <crtl-z>.
Again, scrolling math is impossible to fix in the 1.3.x series.
I agree that this might be annoying but there's really not much that can
be done here.
> 7. In the former version, one could delete the index "a" of "x_a" by
> pressing delete just after the x. Now one has to move into the
> index-box, delete the index there and move out of the index box again.
> That is too involved. In addition, when one only wants to delete x
> and replace it by say y, then it happens very often that one also
> deletes the index, because one wasn't yet in the box.
As the box is only shown when you are inside, it is fairly easy to
figure out whether you're in or out. This might take some time to get
used to, but once you are it is much closer to 'structural editing' than
before.
> Presing <ctrl-z>
> to undo this than confuses you by scrolling the screen...
>
> 8. In the Insert-Cross-Reference-menu, there is no possibility to
> choose the reference with the keybord (apart from writing its name),
> but one has to click with the mouse. In particular the tabulator key
> is not working in many menues to navigate through the various input
> areas. It doesn't hurt, if one uses it only from time to time, but the
> cross references, I'd really like to include via keyboard (as it was
> possible in the old version)
This is a thing that annoys me too...
> 8. in general I would suggest to keep the LyX-input as closely as
> possible to the LaTeX input.
Note however that .tex is not the only output target for LyX and it is
certainly not the most well-structured etc. So in places where there is
a decision between _either_ supporting a structured approach _or_
keeping 1:1 in sync with TeX the decision will most likely fall in
favour of the former.
> I really want to know what I'm doing, and so it's nice, when you
> are writing LaTeX code and LyX just interpretes it. I'm afraid that
> LyX is in some respects (for example by this automatic boxing) leaving
> this path and approaching some windows-like math-editors.
I am trying not to take this as a personal insult ;-)
> Thanks for reading - I'm really a fan of LyX and I hope some of this
> will be of help for you. Best regards, Sebastian Guttenberg
Thanks for the feedback, which is most welcome even if I sound a bit
harsh from time to time.
Andre'
