Re: [GNUnet-developers] Encrypted Message

[Nils Gillmann]

Nils Gillmann transcribed 3.0K bytes:
> I'm responding to the list since I see no reason why this
> should be offlist. I hope this is okay for you.
> 
> Christian Grothoff transcribed 4.4K bytes:
> 
> > >> I will probably finish this today or tomorrow, if it builds then, we need
> > >> a couple more eyes to review the content, types, URLs and so forth.
> > > 
> > > Change of course: I took a 2nd closer look at anonbib and for now it
> > > is inappropriate for the work which would have to put into it for our
> > > bibtex lib. Reasons include: python2 specific issues which take too
> > > long to fix.
> > 
> > That's a very, very brief reason to give for a major change in
> > direction. I don't understand at all what "python2 specific issues"
> > would even be that we'd care about here. Please do elaborate _a lot_.
> > 
> > > I have started with bibtex2html, perl based. This should give us a
> > > html output in short time. I'll work on this and tell you once it is
> > > usable and when you can contribute to the work on it.
> > 
> > Eh, this makes no sense. To start with, Perl is not really a nice
> > language (especially for this). We don't have anything in Perl so far!
> > Also, it _does_ make sense to work with the Tor people on this one, so
> > if there are issues both projects could benefit.
> > 
> > So without further information, I'd be against this.
> > 
> 
> Okay, I've slept over my last emails.
> 
> To be fair, your argument that we don't have anything in perl and that
> perl is not really a nice language is really just your opinion.
> Except for styling we won't change much in the code.[1]
> Anyway, back to my conclusions:
> 
> The whole code of anonbib is in python2. It makes no use of maybe
> more usable and well maintained (or: well written, more general)
> bibtex modules in python. They imported their BibTeX module from
> a perl module (read the code for all the reasons I mention here).
> I don't see myself pushing code in their direction, because I
> would have to change lots of the code base. A non-issue which
> needs to be fixed nevertheless is that they have python2
> exlusively - and we are working on getting rid of python2.
> 
> The issues with python2 in the code:
> It comes down to the choice of either fixing many of our entries to
> their limited BibTeX implementation (which is the main cause of
> errors, we have more valid BibTeX record types), something which in
> some parts can not be done with sed, or extending the python2 code.
> If you haven't read the python2 code, please do. It is really not
> pleasant to read in some parts. Extending will take longer than I can
> justify, which leads me to:
> 
> Last night I looked into jinja2 based bibtex->html. It is easy.  As we
> already use jinja2, this makes more sense than maybe eventually
> sending code to freehaven's anonbib. I'll write a separate website for
> this in jinaj2 and a common used bibtex module. Later we can integrate
> this into the mainwebsite, but it makes sense to keep them separate.
> I hope this is alright, if it isn't someone else can continue with
> anonbib.
> 
> 1: Okay, I just remembered after writing this email that we discussed
>    extensions to this before: we wanted some kind of search and other
>    features.


I am currently going through our .bib file with a newer python application.
This showed more remaining errors (This time: with the linenumber, much faster)
and I am confident that simply using something which makes use of for example
pybtex will be easier to maintain.