Path: ...!news.mixmin.net!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: David Brown <david.brown@hesbynett.no>
Newsgroups: comp.arch
Subject: Re: Microsoft makes a lot of money, Is Intel exceptionally
 unsuccessful as an architecture designer?
Date: Mon, 23 Sep 2024 16:35:02 +0200
Organization: A noiseless patient Spider
Lines: 104
Message-ID: <vcruam$2ocmq$1@dont-email.me>
References: <memo.20240913205156.19028s@jgd.cix.co.uk>
 <vcd7lo$3op6a$2@dont-email.me> <vcdele$293s$1@gal.iecc.com>
 <vcdstq$3vei1$1@dont-email.me> <vcfdld$525$1@gal.iecc.com>
 <vcfu73$9bc9$2@dont-email.me> <vcgiq7$fmaa$3@dont-email.me>
 <87h6ab33p3.fsf@localhost> <vcid1g$ov66$5@dont-email.me>
 <ca3a58a82e35cb494662818a4c8f3e2a@www.novabbs.org>
 <vcj2ln$vddu$2@dont-email.me> <vcm6jn$1igur$1@dont-email.me>
 <vcrflb$2lnor$4@dont-email.me> <vcrgnb$2m4gq$1@dont-email.me>
 <vcrn5j$2n9qd$1@dont-email.me>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Mon, 23 Sep 2024 16:35:03 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="62bd3a602a060bc6951f98aedab86581";
	logging-data="2896602"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX18Fq+PhT8v7E8XrexzrffruydjI/BPXXdU="
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.0
Cancel-Lock: sha1:J68GgV2+r9CNl2kVysp2jq9qMWQ=
Content-Language: en-GB
In-Reply-To: <vcrn5j$2n9qd$1@dont-email.me>
Bytes: 5965

On 23/09/2024 14:32, Thomas Koenig wrote:
> David Brown <david.brown@hesbynett.no> schrieb:
>> On 23/09/2024 12:24, Thomas Koenig wrote:
> 
>>> But LaTeX has its own issues;  typst can actually be a viable
>>> alternative for some of the things that are difficult with LaTeX,
>>> due to its limitations as a macro on top of TeX (separate BibTeX
>>> runs, for example).
>>
>> Sure, LaTeX has its own challenges.  It is not very WYSIWYG, even with
>> tools like LyX.
> 
> I wrote my PhD thesis in LaTeX, a few decades ago (plus I snuck
> in a few research report in LaTeX, made to look like the Word
> templates that were in use at the time).
> 
>> Macro programming for LaTeX is not always the simplest
>> task (though there are packages that make it a lot easier).
> 
> I always stayed away from that.

There are fun things that you can do with it.  Most of the macros I 
write are very simple, such as:

\newcommand{\uint}[1]{\textbf{uint$_{#1}$}}

Then I can document a field in a network packet as type \uint{32} with 
nice formatting, without the details of the formatting being part of the 
text I am writing.  Sometimes these kinds of macros are more complex, 
with cross-references, or multiple parameters that are put together in a 
more complex layout, and so on.

But some of the more fun stuff I've done involves moving and copying 
data around.  The main parts of the text might describe message types, 
laid out in a standardised format.  An appendix could contain a summary 
table of those message types, generated automatically.

It can, however, get quite messy fairly quickly - and then I am more 
likely to write a Python script that generates the LaTeX code to be 
included with the rest of the document.  These days I use LuaLaTeX, 
which supports writing macros in Lua, but as I have little experience 
with Lua it isn't that much help.  (LuaLaTeX has lots of other nice 
features and packages.)

Long ago, I did some metafont and metapost work too.  These were also 
rather interesting languages - it's the only language I know that has a 
built-in equation solver (for sets of linear equations and Bezier 
curves).  You can write "8 - x = 2 * y; 3 * x + y = 9;" just as easily 
as "x = 2; y = 3;".

> 
>> And it's
>> not hard to make a small typo with the result of reams of
>> incomprehensible errors instead of nice document.
> 
> Been there, done that :-)

I expect it's inevitable for any Turing complete language.  Now there's 
an idea for a thesis for any computational theory students - prove that 
conjecture true or false!

> 
>> But you can also do a
>> lot of nice stuff with it, and produce very high quality results.  And
>> unlike word processors or online systems like typst, it's easy to
>> generate combined sources for LaTeX from other tools -
> 
> I didn't use (and don't particularly recommend) the online, I
> use the one that you invoke very much like LaTeX, just use your
> favorite text editor and issue the command "typst complile foo.typ".

OK.  The pricing page gave that as "contact sales", as far as I could 
see, which hugely increases the bureaucratic burden.  But it also seems 
to be easily downloadable as open source from github.  Is there a 
difference in these?

Handling everything locally makes it very much more attractive to me, 
and would work for mixing generated text with template code - something 
I need and can do with LaTeX but cannot do with online systems.

> 
> And because typst actually has a fairly decend command language, many
> things that would be torture to write in LaTeX are much easier.
> 
> And being able to write a/b instead of \frac{a}{b} _is_ much nicer.
> 

I can see that being nicer, yes.


>> I've put together
>> systems using templates, data from logger systems, databases, generated
>> graphs, etc., controlled by Python code to make automatic reports that
>> are well beyond what can be done with many other tools.
> 
> That, you could do as well with typst.  It even contains a decent
> scripting language, see https://typst.app/docs/reference/scripting/ .

You have convinced me that it is worth investigating.  I'll need to 
compare the features it has to what I have used in LaTeX, and the 
quality of the output.  I greatly appreciate the recommendation!