CTAN submission -- Update of l2tabu
Reinhard Zierke
I have uploaded `l2tabu-1.6.tar.gz' (l2tabu.{pdf,tex}, Version 1.6)
to ftp.dante.de/incoming/
Name of contribution: Das LaTeX-Sündenregister (sins of LaTeX users)
Suggested location on CTAN: info/l2tabu/german/
Summary description: Obsolete commands, packages and other common
mistakes
License type: Free (Open Publication License v1.0)
Announcement text:
----------------------------------------------------------------------
A small list of obsoleted commands and packages that are constantly
posted in "minimal examples" to de.comp.text.tex. I further listed
common mistakes (e.g. \begin{appendix}...\end{appendix} instead of
\appendix ...). This file should be considered an add-on to the
available documentation like l2kurz and FAQs.
This document is written in German.
Changes: * Added \sloppy
* Added section numbers to FAQ references (changed FAQ Version
to v72)
* Some minor corrections
----------------------------------------------------------------------
Please untar the archive to CTAN:info/l2tabu/german/
----- End forwarded message -----
Thanks for the upload. I installed it as suggested replacing the old
version in CTAN:info/l2tabu/german/ (It moved there a short while ago
from CTAN:info/german/l2tabu/ when an English translation was established
in info/l2tabu/english/)
Reinhard Zierke
for the CTAN team
Donald Arseneau
> A small list of obsoleted commands and packages that are constantly
>
> Changes: * Added \sloppy
I hope this means the document is formatted with \sloppy, and
not that \sloppy has been listed as obsolete!
Donald Arseneau as...@triumf.ca
Harald Harders
> zie...@dante.de (Reinhard Zierke) writes:
>
>> A small list of obsoleted commands and packages that are constantly
>>
>> Changes: * Added \sloppy
>
> I hope this means the document is formatted with \sloppy, and
> not that \sloppy has been listed as obsolete!
I contrast, I hope this does not mean the document is formatted
with \sloppy. I hope it is written that you should use \sloppy
only if really necessary. l2tabu does not only speak about
obsolete things but also about other things you should avoid
if possible.
Yours
Harald
--
Harald Harders Langer Kamp 8
Institut für Werkstoffe D-38106 Braunschweig
Technische Universität Braunschweig Germany
E-Mail: h.ha...@tu-bs.de Tel: +49 (5 31) 3 91-3062
WWW : http://www.tu-bs.de/institute/ifw/ Fax: +49 (5 31) 3 91-3058
Mark Trettin
* Donald Arseneau <as...@triumf.ca> wrote:
> zie...@dante.de (Reinhard Zierke) writes:
>
>> A small list of obsoleted commands and packages that are constantly
>>
>> Changes: * Added \sloppy
>
> I hope this means the document is formatted with \sloppy, and
> not that \sloppy has been listed as obsolete!
Neither. It's listed as something that one should avoid as long as
possible. The problem is that a lot of people still don't you use
T1-encoded fonts and therfore words with umlauts don't get hyphenated
well.
Greetings
Mark
--
Mark Trettin · Aachen · Germany · Where is Aachen? --> N: 50°46' E: 06°05'
BOFH excuse #199:
the curls in your keyboard cord are losing electricity.
Donald Arseneau
> Neither. It's listed as something that one should avoid as long as
> possible. The problem is that a lot of people still don't you use
> T1-encoded fonts and therfore words with umlauts don't get hyphenated
> well.
#2, fine; #1, wrong. Every document should be formatted with \sloppy
unless cpu speed is a major impediment. \sloppy can only help; it never
hurts. The extra line-break possibilities allowed by \sloppy are always
better then the alternate line breaks. TeX will still warn of underfull
boxes when no decent break was found.
LaTeX's [draft] option is really stupid as it sets \overfullrule *and*
\sloppy! \overfullrule might be useful without \sloppy.
Donald Arseneau as...@triumf.ca
William F. Adams
>#2, fine; #1, wrong. Every document should be formatted with \sloppy
>unless cpu speed is a major impediment. \sloppy can only help; it never
>hurts. The extra line-break possibilities allowed by \sloppy are always
>better then the alternate line breaks. TeX will still warn of underfull
>boxes when no decent break was found.
Hmm, this is interesting.
I will note that \sloppy will yield IME different breaks than
\begin/end{sloppypar}, and on occasion will leave stuff hanging out in the
margin (I believe this supports Donald's assertion that it won't things get as
ugly as sloppypar will).
William
--
William Adams
http://members.aol.com/willadams
Sphinx of black quartz, judge my vow.
Donald Arseneau
> Hmm, this is interesting.
>
> I will note that \sloppy will yield IME different breaks than
> \begin/end{sloppypar}, and on occasion will leave stuff hanging out in the
> margin
Oooooo. That *is* interesting. Really?
> (I believe this supports Donald's assertion that it won't things get as
> ugly as sloppypar will).
But I don't assert that. What I do assert is that \sloppy eliminates (most)
"hanging out in the margin". When the linebreaks are already good without
\sloppy, they will be just as good (usually the same) with \sloppy.
Donald Arseneau as...@triumf.ca
Walter Schmidt
>
> When the linebreaks are already good without
> \sloppy, they will be just as good (usually the same) with \sloppy.
This is my experience, too, but when I discussed this subject
some time ago on the german mail list TEX-D-L, a number of
experts claimed it was not true. Rationale (simplified):
With \sloppy, the line breaking algorithm may terminate
already with a "sloppy" result before finding the optimum.
I do not have the kowledge to prove that this reasoning is
wrong, yet I don't believe it. I'm glad to hear that you
back my opinion :-)
best wishes
Walter
Brian Blackmore
Donald is correct about this. As discussed on page 96 of The TeXbook,
line breaking is done in three passes. The first pass makes attempts
without discretionary hyphens, and this pass succeeds "if... none of
the... lines has badness exceeding... \pretolerance". As a second
attempt, hyphenation is allowed and "the attempt is made using \tolerance
instead of \pretolerance".
Since \sloppy doesn't change \pretolerance, the first-pass behavior will
be identical with or without \sloppy. For the next pass, LaTeX's default
\tolerance is 200 (this is also the setting for \fussy), but \sloppy uses
a \tolerance of 9999. That means that any line breakage that is accepted
without \sloppy will be accepted with \sloppy. Here, however, is where
\sloppy allows for more line-breaking possibilities before forcing TeX to
exercise its third, and final option for breaking lines.
It should also be noted that \sloppy increases \hfuzz, so more line
badness will be allowed before an overfull hbox is reported.
--
Brian Blackmore
blb8 at po dot cwru dot edu
Torsten Bronger
Brian Blackmore <bl...@po.cwru.edu> writes:
> [...]
>
> Since \sloppy doesn't change \pretolerance, the first-pass behavior will
> be identical with or without \sloppy. For the next pass, LaTeX's default
> \tolerance is 200 (this is also the setting for \fussy), but \sloppy uses
> a \tolerance of 9999. That means that any line breakage that is accepted
> without \sloppy will be accepted with \sloppy. Here, however, is where
> \sloppy allows for more line-breaking possibilities before forcing TeX to
> exercise its third, and final option for breaking lines.
So, if I include a local \sloppy for every paragraph that caused an
overfull hbox, this is equivalent to one global \sloppy?
Tschoe,
Torsten.
--
Torsten Bronger, aquisgrana, europa vetus
Brian Blackmore
No, because of:
Markus Kohm
> That means that any line breakage that is accepted
> without \sloppy will be accepted with \sloppy.
Only for those paragraphs without hyphenations. But if you have hyphenation
at the paragraph, \tolerance was used for the paragraph and line breakage
may be not as good as it could be.
Proof:
\documentclass{article}
\setlength{\textwidth}{20em}
\setlength{\parindent}{0pt}
\begin{document}
\typeout{First without \string\sloppy\space and underfull \string\hbox}
tatata tatata tatata tatata tatata tatata ta\-ta\-tata
tatata tatata tatata tatata tatata tatata tata\-tata
tatata tatata tatata tatata ta\-tatatatt\-ta
tatata tatata tatata tatata tatata tatata ta\-ta\-ta\-ta
\typeout{done.}
\sloppy
\typeout{Second with \string\sloppy\space and underfull \string\hbox}
tatata tatata tatata tatata tatata tatata ta\-ta\-tata
tatata tatata tatata tatata tatata tatata tata\-tata
tatata tatata tatata tatata ta\-tatatatt\-ta
tatata tatata tatata tatata tatata tatata ta\-ta\-ta\-ta
\typeout{done.}
\end{document}
Markus
Donald Arseneau
> So, if I include a local \sloppy for every paragraph that caused an
> overfull hbox, this is equivalent to one global \sloppy?
Yes. (I do have to hedge because \sloppy *can* choose different
line breaks from the normal fussy setting, even when there are no
overfull boxes, but it is rather rare.)
The point about increased \hfuzz is misplaced because it only
applies when there would be overfull boxes under \fussy.
But increased \hfuzz *is* a valid counter-argument to my original
assertion that \sloppy has no downside.
Donald Arseneau as...@triumf.ca
Donald Arseneau
> Brian Blackmore wrote:
>
> > That means that any line breakage that is accepted
> > without \sloppy will be accepted with \sloppy.
For some meaning of "accepted" ... Overfull lines (even those within
\hfuzz, alas) are not "accepted", but rather "forced".
> Only for those paragraphs without hyphenations. But if you have hyphenation
> at the paragraph, \tolerance was used for the paragraph and line breakage
> may be not as good as it could be.
In such cases, if you prefer the line-breaking under \fussy, with
multiple consecutive hyphenations, then you should have lower values
of \doublehyphendemerits and \finalhyphendemerits.
You are right that \tolerance only applies to badness, not penalties,
so that, with low \tolerance, Tex will choose very disfavored (highly
penalized) line-breaks rather than stretch spaces a modest amount.
> Proof:
Surprisingly, no. I know it looks like it is a case of \doublehyphendemerits
and \finalhyphendemerits, but the real difference is that the first
fussy "good" paragraph actually has an overfull hbox!
Donald Arseneau as...@triumf.ca
Markus Kohm
> I know it looks like it is a case of \doublehyphendemerits
> and \finalhyphendemerits, but the real difference is that the first
> fussy "good" paragraph actually has an overfull hbox!
But a very, very small (0.00021pt which is less than 0.1µm), which is
tollerated by the \fussy settings (no overfull \hbox message) and nobody
would see it. But the streching of the \sloppy version is very easy
recognizable.
Because of this I think \sloppy parbreaking may be not as good as \fuzzy if
\fuzzy doesn't message overfull \hboxes. Everybody who agrees shouldn't use
\sloppy in general but sloppypar if needed.
Markus
Walter Schmidt
>
> Donald Arseneau wrote:
>
> > I know it looks like it is a case of \doublehyphendemerits
> > and \finalhyphendemerits, but the real difference is that the first
> > fussy "good" paragraph actually has an overfull hbox!
>
> tollerated by the \fussy settings (no overfull \hbox message) and nobody
> would see it. But the streching of the \sloppy version is very easy
> recognizable.
This means: With \sloppy turned on, sloppy paragraph breaking
(i.e. the third "run") is performed as soon as there is an
overfull hbox after the second run, regardless of how small the
overflow is? With other words, \hfuzz is ignored with \sloppy?
This looks like a very unfortunate feature of TeX's paragraph
breaking algorithm (unless I misunderstand what's going on).
IMHO, a better approach would be to attempt sloppy paragraph
breaking only if \hfuzz is exceeded. (What about fixing this
in ExTeX, Markus?)
just my 2c
Walter
Dan Luecking
wrote:
>Markus Kohm schrieb:
>>
>> Donald Arseneau wrote:
>>
>> > I know it looks like it is a case of \doublehyphendemerits
>> > and \finalhyphendemerits, but the real difference is that the first
>> > fussy "good" paragraph actually has an overfull hbox!
>>
>> But a very, very small (0.00021pt which is less than 0.1痠), which is
>> tollerated by the \fussy settings (no overfull \hbox message) and nobody
>> would see it. But the streching of the \sloppy version is very easy
>> recognizable.
>
>This means: With \sloppy turned on, sloppy paragraph breaking
>(i.e. the third "run") is performed as soon as there is an
>overfull hbox after the second run, regardless of how small the
>overflow is? With other words, \hfuzz is ignored with \sloppy?
I think (and I will surely be corrected if wrong) that \hfuzz is
_always_ ignored by the line breaking algorithm. It only affects
the _reporting_ of overfull boxes when the line breaking is
completed. TeX still tries to have no amount of overfullness at
all.
>
>This looks like a very unfortunate feature of TeX's paragraph
>breaking algorithm (unless I misunderstand what's going on).
>IMHO, a better approach would be to attempt sloppy paragraph
>breaking only if \hfuzz is exceeded.
The third run (using \emergencystretch) was added to TeX later. I
agree that TeX should not attempt it unless \hfuzz has been exceeded.
But that is not under the control of \sloppy or, I think, any
parameter one can set.
> (What about fixing this
>in ExTeX, Markus?)
If you mean etex, then that could be done, but if you meant EmTeX, then
that would make it not TeX.
Dan
--
Dan Luecking Department of Mathematical Sciences
University of Arkansas Fayetteville, Arkansas 72701
luecking at uark dot edu
Markus Kohm
> With other words, \hfuzz is ignored with \sloppy?
More or less \hfuzz is only "report overfull \hboxes greater than this
value" and not "avoid overfull \hboxes greater than this value".
> What about fixing this in ExTeX, Markus?
Not in gerneral but as an alternative, this would be a good idea. There may
be several alternative parbreak algorithms at ExTeX - and more parameters
for the existing one. We should remember this alternative, when we'll
implement the par-typesetter.
Markus