[tex-live] tl-pretest: bug in .ini files

[Petr Olsak]

On Thu, 14 Apr 2016, Norbert Preining wrote:

>> This is totally unfair.
>
> Unfair? Maybe call it development? We have changed several
> configuration file formats a few times, due to needs and
> convenience.

Yes. Almost all such change brought problems in the past. But I don't want 
to return to them.

We expect a development in (relatively) new sofware like LuaTeX, XeTeX. 
But pdfTeX is very old and many people use it for its (relatively) 
stability as an engine with direct output to PDF and with microtypographic 
enhancements. And changing the core .tex file used in .ini files for 
pdfTeX is a bad message: "no, pdfTeX is not stable".

> We don't do these kind of things lightly, but it happens.
>
> Nothing *unfair* here.

It *is unfair* when you are doing it in (relatively) stable software.

>> Please, keep the old format of pdftexconfig.tex. You cannot imagine
>> how many automatons exist in various linux computers based on the
>> old format of pdftexconfig.tex.
>
> Let us know how many? My guess is *1*, yours.

No only one. There are manuals on internet which describe internals of 
TeXlive configuration. People use them. For example the discussion about 
"tlmgr paper" a few days ago is *only one* example of many others similar 
problems. Do you know how many linux distributions use pdftexconfig.tex 
directly in other manner than how it is used in TeXlive?

>> "pdftex -ini" is default starting point for pdfTeX and this default
>
> In TL we are having
> 	pdftex -ini ... '*pdfetex.ini'
> which activates the eTeX extensions.
>
> And that is the fact since several years AFAIR.

"*pdftex.ini" is only one way how to generate pdftex which is based on 
etex.src. People can use many other ways, for example without etex.src.

If fact, there is -enc extension in pdftex (since 2000, i.e. "several 
years") and I don't expect that "it is pretty unusual to want no encTeX".

It is evident that there is a bug in pdftexconfig.ini because \unless is 
used here. I hope that this needs no more discussion.

>> does not indludes eTeX extensions. You cannot generalize what is
>> "prety unusual" only from the state of your computer.
>
> And *you* cannot generalize what is usual from *your* computer's state.

Sorry, this is not only my copmuter's state. People, for example, use my 
books where behavior of software is described. When it is changed by 
blusterous developers then this part of the book gets to be invalid. The 
last case of this type was the changing behavior of --no-virtual option in 
the otftotfm. Kerns and ligatures were generated but newly they are not 
generated. Unless exactly we need not to use virtual fonts for implementig 
this. And my new book (TeX pro pragmatiky) gets to be invalid on page 132. 
The *printed* form of this book were sent to hundreds TeX users:(.

This is reason why I am very deliberate during each "development" change.

Petr