Bug in tabu with scantokens (GL, please fix)

[blfl...@gmail.com]

Hi GL,

please see

http://tex.stackexchange.com/questions/87144/

for a bug report about tabu. Basically, tabu does

\scantokens{\def\:{...}}

when it should do

\everyeof{\noexpand}\edef\:{\scantokens{...}}

(that's within a group anyway). Well, actually the correct fix is a bit more elaborate, and it is not clear that using \scantokens is correct at all. See question, answer, and comments there.


Bruno

[GL]

Le 20/01/2013 01:58, blfl...@gmail.com a �crit :

> Hi GL,
>
> please see
>
> http://tex.stackexchange.com/questions/87144/
>
> for a bug report about tabu. Basically, tabu does
>
> \scantokens{\def\:{...}}
>
> when it should do
>
> \everyeof{\noexpand}\edef\:{\scantokens{...}}

And please why ?

>
> (that's within a group anyway). Well, actually the correct fix is a bit more elaborate, and it is not clear that using \scantokens is correct at all. See question, answer, and comments there.

When you use \scantokens, you will always find dozens of people
to say that it's a "Danger". imo \catcode scares them !

Well, to be honest, the time of bug reports will come with the next
release. I'm just implementing the \multirow feature, and I have to
say that the current available code is very far from what i'm working
on now.


I've looked at the example and i've no such problem (with my version).

But {tabu*} fails in this case... and i'll investigate.


fyi: tabu collects it's environment contents. So \scantokens
is compulsory when inserting verbatim material.


Regards.

[Enrico Gregorio]

GL <goua...@gmail.com> wrote:

> Le 20/01/2013 01:58, blfl...@gmail.com a ÔøΩcrit :

> > Hi GL,
> >
> > please see
> >
> > http://tex.stackexchange.com/questions/87144/
> >
> > for a bug report about tabu. Basically, tabu does
> >
> > \scantokens{\def\:{...}}
> >
> > when it should do
> >
> > \everyeof{\noexpand}\edef\:{\scantokens{...}}
>
> And please why ?

Making people being able to use your package trusting in
its documentation isn't a good motivation?

> > (that's within a group anyway). Well, actually the correct fix is a bit
> > more elaborate, and it is not clear that using \scantokens is correct at
> > all. See question, answer, and comments there.
>
> When you use \scantokens, you will always find dozens of people
> to say that it's a "Danger". imo \catcode scares them !

But if /you/ use it, please ensure it does the right thing.

> Well, to be honest, the time of bug reports will come with the next
> release. I'm just implementing the \multirow feature, and I have to
> say that the current available code is very far from what i'm working
> on now.

So you're telling that the current version of tabu is
completely unmaintained?

> I've looked at the example and i've no such problem (with my version).

Which is /not/ the version common people have.

> [...]

Ciao
Enrico

[GL]

Le 20/01/2013 11:20, Enrico Gregorio a ÔøΩcrit :

> GL <goua...@gmail.com> wrote:
>
> So you're telling that the current version of tabu is
> completely unmaintained?

In three letters, with no ligature: yes !

And if it displease you:
1) I'm working alone
2) I'm not paid for that
3) Less than 5% of the commands are the same between the
next version and the uploaded one
4) I have to finish \multirow (minimum 3 weeks)
5) I have to implement a longtabu environment (unknown time)

The example shown in the link does not make use
of any feature of tabu. The OP can use tabularx
for that.

Ciao-ciao.

[Enrico Gregorio]

GL <goua...@gmail.com> wrote:

> Le 20/01/2013 11:20, Enrico Gregorio a ÔøΩcrit :
> > GL <goua...@gmail.com> wrote:
> >
> > So you're telling that the current version of tabu is
> > completely unmaintained?
>
> In three letters, with no ligature: yes !

It would be good that you informed the possible users
of your package; just start the doc of a new version
with

The present version of tabu is unmaintained
and no bug reports about it will be answered

> And if it displease you:
> 1) I'm working alone

I've got no team at my disposal for my packages.

> 2) I'm not paid for that

Very few, if any, among the package authors are.

> 3) Less than 5% of the commands are the same between the
> next version and the uploaded one

I hope not the user's interface commands.

> 4) I have to finish \multirow (minimum 3 weeks)
> 5) I have to implement a longtabu environment (unknown time)

I sincerely hope you'll solve your problems.

> The example shown in the link does not make use
> of any feature of tabu. The OP can use tabularx
> for that.

I beg to differ: \begin{tabu}to\textwidth{|l|X[-1,l]|}
uses specific tabu features.

Ciao
Enrico

[GL]

Le 20/01/2013 15:07, Enrico Gregorio a écrit :

> GL <goua...@gmail.com> wrote:
>
>> 3) Less than 5% of the commands are the same between the
>> next version and the uploaded one
>
> I hope not the user's interface commands.

Of course yes. Everything has been rewritten.

>> 4) I have to finish \multirow (minimum 3 weeks)
>> 5) I have to implement a longtabu environment (unknown time)
>
> I sincerely hope you'll solve your problems.

A lot of. For \multirow, i think i have the method, but
the implementation's tricky.

For longtabu, i want to be able to use it inside
multicols, thus everything has to be done (you can think
about it ;-) )

>> The example shown in the link does not make use
>> of any feature of tabu. The OP can use tabularx
>> for that.
>
> I beg to differ: \begin{tabu}to\textwidth{|l|X[-1,l]|}
> uses specific tabu features.

Not at all: X[-1] alone is exactly the same as X[1] thus the
same as tabularx X-column.

What did you expect it meant ?
(I'm interested in the answer, to check if the interface
is correct, for the next release is different)


>
> Ciao
> Enrico
>

[Enrico Gregorio]

GL <goua...@gmail.com> wrote:

> Le 20/01/2013 15:07, Enrico Gregorio a ÔøΩcrit :

> > GL <goua...@gmail.com> wrote:
> >
> >> 3) Less than 5% of the commands are the same between the
> >> next version and the uploaded one
> >
> > I hope not the user's interface commands.
>
> Of course yes. Everything has been rewritten.

So old documents written with the previous commands
of tabu will cease to work? Wonderful.

> [...]

> > I beg to differ: \begin{tabu}to\textwidth{|l|X[-1,l]|}
> > uses specific tabu features.
>
> Not at all: X[-1] alone is exactly the same as X[1] thus the
> same as tabularx X-column.
>
> What did you expect it meant ?
> (I'm interested in the answer, to check if the interface
> is correct, for the next release is different)

I don't know, as I /never/ use tabu. I tried some times,
but the documentation is unreadable. That specification is

X[-1,l]

that seems quite different from X[1] and X[-1].

Ciao
Enrico

[GL]

Le 20/01/2013 16:45, Enrico Gregorio a ÔøΩcrit :

> GL <goua...@gmail.com> wrote:
>
>> Le 20/01/2013 15:07, Enrico Gregorio a ÔøΩcrit :
>>> GL <goua...@gmail.com> wrote:
>>>
>>>> 3) Less than 5% of the commands are the same between the
>>>> next version and the uploaded one
>>>
>>> I hope not the user's interface commands.
>>
>> Of course yes. Everything has been rewritten.
>
> So old documents written with the previous commands
> of tabu will cease to work? Wonderful.

I don't care.
I want to built tabulars without any measurement or guess,
with a maximum of features and a powerful interface.

The next doc will be shorter, and divided into different
documents (with hyperlinks of course). One of them will
provide a lot of examples (50 or so). Examples are better
to get started.

[Enrico Gregorio]

GL <goua...@gmail.com> wrote:

> Le 20/01/2013 16:45, Enrico Gregorio a ÔøΩcrit :

> [...]

> > So old documents written with the previous commands
> > of tabu will cease to work? Wonderful.
>
> I don't care.

Users do.

> I want to built tabulars without any measurement or guess,
> with a maximum of features and a powerful interface.
>
> The next doc will be shorter, and divided into different
> documents (with hyperlinks of course). One of them will
> provide a lot of examples (50 or so). Examples are better
> to get started.

As long as the new package has a new name, all's well.
If it has the same name, then many people won't be happy
and will have to choose between installing the new
version or keeping the old one.

It doesn't matter how many bells and whistles the new
version will have: people want that their document
compile to begin with. It can happen that a new version
of some software introduces some small back incompatibility
(with suggestions on how to fix it). A new version that
completely changes the interface is unacceptable.

Ciao
Enrico

[GL]

Le 20/01/2013 17:33, Enrico Gregorio a ÔøΩcrit :

> GL <goua...@gmail.com> wrote:
>
>> Le 20/01/2013 16:45, Enrico Gregorio a ÔøΩcrit :
>> [...]
>>> So old documents written with the previous commands
>>> of tabu will cease to work? Wonderful.
>>
>> I don't care.
>
> Users do.

The more funny is that it's you, Enrico, who is (wisely) not using tabu,
who complains the most !

Good evening.

[Bernhard Tempel]

Am 20.01.2013 19:00, schrieb GL:

> Le 20/01/2013 17:33, Enrico Gregorio a écrit :
>> GL <goua...@gmail.com> wrote:
>>

>>> Le 20/01/2013 16:45, Enrico Gregorio a ⁄crit :

>>> [...]
>>>> So old documents written with the previous commands
>>>> of tabu will cease to work? Wonderful.
>>>
>>> I don't care.
>>
>> Users do.
>
> The more funny is that it's you, Enrico, who is (wisely) not using tabu,
> who complains the most !

Maybe he is at the moment. But others are reading this conversation. And well,
to be honest, I - as a user - consider this conversation, especially your "I
don't care" as a strong warning not to rely on tabu for any important document.

Have a nice evening,

Bernhard

[Axel Berger]

Bernhard Tempel wrote:
> a strong warning not to rely on tabu for any important document.

It seems Markus Kohm has hit the same problem more than once, which is
why we now have scrlettr.cls and scrlttr2.cls as well as scrpage.sty and
scrpage2.sty. I really recommend his exemplary way to go about things
at, one may add, no extra effort and very little inconvenience.

Axel

[GL]

That's right.
The points are that:
- colortbl does not satisfy me (too many reasons...)
- booktabs rules are nice but incompatible with vertical leaders
and cell colouring (unless heavy endevours...)
- siunitx alignments too often makes too wide columns or numbers
overtake the cell borders
- vertical spacing is too often bad with font change / math inclusion
- multicolumn requires extra measurement (what dimen for p{...} ?)
- multirow requires extra measurement as well and gives acceptable
results only in simple cases.
- pgf provides a lot of drawings / fadings are would be nice to
include in a table quite easily.

Presently for typesetting tabulars with robustness you get:
. array.sty
. tabularx.sty
. booktabs.sty
. longtable.sty
All those are very good and will never be superseeded by tabu.
Any other enhancement causes problems (including colortbl).

That's why tabu is not designed as an enhancement but more as
a /bundle/. That's why it's long to develop.

>
> Have a nice evening,

You too. (Snow over Paris this weekend ;-) )

>
> Bernhard
>

[Steffen Hoffmann]

GL schrieb am 20.01.2013:

> Le 20/01/2013 17:33, Enrico Gregorio a écrit :
>> GL <goua...@gmail.com> wrote:

>>> Le 20/01/2013 16:45, Enrico Gregorio a ⁄crit :

>>> [...]
>>>> So old documents written with the previous commands
>>>> of tabu will cease to work? Wonderful.
>>> I don't care.
>> Users do.
> The more funny is that it's you, Enrico, who is (wisely) not using tabu,
> who complains the most !

No, he was just for some the only one, who replied to you!

And he does user support, and I guess, he anticipates lots of user
complaints.


It is, of course, your choice, whether you want to develope the
consisting code or rather start a new version from scratch.

But in the second case you *should* provide a solution for users of
legacy versions.

You were already pointed to KOMA-Script, where Markus Kohm decided at
some point to make a new start for the letter class and the page heading
package – and he decided to leave alone the legacy versions and gave the
new ones new names.

Another solution with only a little inconvenience for users would be to
rename the legacy package to “oldtabu.sty” / “tabuv28.sty” / whatever or
an approach like in “csquotes” with a package option for older versions.

sph � �Speravir�

[GL]

Le 21/01/2013 23:48, Steffen Hoffmann a écrit :

> GL schrieb am 20.01.2013:
>
> You were already pointed to KOMA-Script, where Markus Kohm decided at
> some point to make a new start for the letter class and the page heading
> package – and he decided to leave alone the legacy versions and gave the
> new ones new names.

Well: to be honest, i'm not developing KOMA-Script and my name is not
Markus Kohm.

> Another solution with only a little inconvenience for users would be to
> rename the legacy package to “oldtabu.sty” / “tabuv28.sty” / whatever or
> an approach like in “csquotes” with a package option for older versions.

The uploaded version is deceiving. The algorithms are not so correct.
CTAN has pleinty of packages that cause problems. The new version will
replace the old one and this thread is closed for me.

[Robin Fairbairns]

perhaps we should have put it on macros/latex/exptl in the first case.

shall i delete the existing one from ctan?
--
Robin Fairbairns, Cambridge
sorry about all this posting. i'll go back to sleep in a bit.

[GL]

Le 22/01/2013 00:36, Robin Fairbairns a �crit :
> GL <goua...@gmail.com> writes:
>
>> Le 21/01/2013 23:48, Steffen Hoffmann a �crit :

>>> GL schrieb am 20.01.2013:
>>>
>>> You were already pointed to KOMA-Script, where Markus Kohm decided at
>>> some point to make a new start for the letter class and the page heading

>>> package � and he decided to leave alone the legacy versions and gave the

>>> new ones new names.
>>
>> Well: to be honest, i'm not developing KOMA-Script and my name is not
>> Markus Kohm.
>>
>>> Another solution with only a little inconvenience for users would be to

>>> rename the legacy package to �oldtabu.sty� / �tabuv28.sty� / whatever or
>>> an approach like in �csquotes� with a package option for older versions.

>>
>> The uploaded version is deceiving. The algorithms are not so correct.
>> CTAN has pleinty of packages that cause problems. The new version will
>> replace the old one and this thread is closed for me.
>
> perhaps we should have put it on macros/latex/exptl in the first case.
>
> shall i delete the existing one from ctan?
>

Not before the next one replaces the old one. (will you delete
keycommand, boolexpr and etextools ? three packages that are
worse than the current available TabU version ?... )

v2.9 will come: just be patient. Discussions about TabU are showing
that quite a lot of people is interested in such a package.

[Guenter Milde]

On 2013-01-22, GL wrote:

> Le 22/01/2013 00:36, Robin Fairbairns a écrit :
>> GL <goua...@gmail.com> writes:

>>> Le 21/01/2013 23:48, Steffen Hoffmann a écrit :
>>>> GL schrieb am 20.01.2013:

...

>> perhaps we should have put it on macros/latex/exptl in the first case.

>> shall i delete the existing one from ctan?

> Not before the next one replaces the old one. (will you delete
> keycommand, boolexpr and etextools ? three packages that are
> worse than the current available TabU version ?... )

> v2.9 will come: just be patient. Discussions about TabU are showing
> that quite a lot of people is interested in such a package.

This is why people (even the ones not currently using tabu) care for
consistency. A lot of people are interested in such a package and will feel
misguided if things that more or less work with tabu now will no longer work
without extensive rewriting in their documents. And their number will
increase if the old and not-to-be-continued version stays on on CTAN.

Günter

[1g2...@gmail.com]

> v2.9 will come: just be patient. Discussions about TabU are showing
>
> that quite a lot of people is interested in such a package.

What happened to v2.9? Can we still expect a new version?

[sergio.c...@gmail.com]

Please, do not publish in CTAN something /with the same name/ and a completely different functionality. This will break tons of documents including all those stored in source LaTeX on arXiv.

[GL]

Yes i'm working quite hard on it for the last few months.
I've still some extra work to do on the code, and i've experimented
that coding in TeX is a slow process !

But nothing appears to stand across the road for the new version,
and i would be very happy to deliver it this summer.

As usual, i recommand not to use the present version for important
documents, for the syntax changes in many respect.

Regards.

---
L'absence de virus dans ce courrier électronique a été vérifiée par le logiciel antivirus Avast.
http://www.avast.com

[schrodi...@me.com]

> Yes i'm working quite hard on it for the last few months.
> I've still some extra work to do on the code, and i've experimented
> that coding in TeX is a slow process !
>
> But nothing appears to stand across the road for the new version,
> and i would be very happy to deliver it this summer.
>
> As usual, i recommand not to use the present version for important
> documents, for the syntax changes in many respect.
>
> Regards.
>
> ---
> L'absence de virus dans ce courrier électronique a été vérifiée par le logiciel antivirus Avast.
> http://www.avast.com

Hello, any updates regarding new tabu release? Very much looking forward to it!