[Debian-med-packaging] Bug#761942: [RFR] templates://arb/{arb-common.templates}

[Elmar Pruesse]

As a disclaimer up front: I'm actually part of upstream. As a long term
Debian user I couldn't refuse the honor when Andreas asked me to help
with the package. *Thank you all for the great work you do to make
Debian the most awesome software environment on the planet!*

DebConf Template:

"PT server" is probably best regarded as a name. It's meaning to the
user community is really "something I need to do my analysis".

Technically, it is essentially a search engine for DNA sequence data.
The acronym "PT" is historical and spells indeed "positional tree" or
"prefix tree". Today the data structure is known as "suffix trie". Due
to the lengthy index computation process and the high memory
requirements, the engine runs as a "server process" in the background.
Launch, termination and index calculation are managed from within ARB.
For this, several "slots" comprising a location for index data and a
socket (unix or tcp, optionally remote) can be configured in
arb_tcp.dat. These slots are presented in the GUI and are what the users
see as "PT server".

The users do not distinguish between the program, the running instance,
the configuration slot or the index data. I believe the least confusing
description in the template would be "PT server slot", perhaps with
quotes around "PT server".

> http://help.arb-home.de/ says PT is "Prefix Tree" (or occasionally
> "Positional Tree", but usually the former).  It seems to be
> canonically either "PT_SERVER" or hyphenated "PT-server".

I'd prefer the latter as the human readable form. The former probably
originates from C code.

> Given that the .dat file is under /etc/ I assume "preconfigured" means
> "configured by default" rather than "hardwired".  So:

Precisely.

>     The default configuration in /etc/arb/arb_tcp.dat gives ARB three
>     private Prefix-Tree (PT) servers for each user as well as three
>     global PT-servers accessible by all users.

How about:

"""
The default configuration in /etc/arb/arb_tcp.dat gives ARB three
private PT-server slots for each user as well as three global PT-server
slots accessible by all users.
"""

> The part I don't understand is, does this really mean to say that ARB
> will constantly run 3 communal PT-servers *plus* 3n private servers
> where n equals the total number of local users on the system?  That
> sounds like it could be an awful lot of processes.  Or is the "for
> each user" a mistranslation?  Maybe it really means something more
> like a *maximum* of 3 communal servers plus a *maximum* of 3
> non-communal servers which (as it goes on to say) can only be
> activated by members of the arb group?

Prior to ARB 6.0, the administrator would configure one or more "server
slots" for each user on specific TCP ports. As of 6.0, three slots are
configured to use Unix sockets and data locations within the users home
directory. They are only launched upon user request and terminated when
ARB is closed.


Name (ARB vs Arb):
Publications in molecular biology generally use capitalization to
indicate that a word refers to a software. Tool names are typically
acronyms artfully crafted to resemble actual words (e.g. BLAST, MUSCLE)
and lower case would often lead to confusion or even unintended puns. I
could make "Arb" the official name to reduce FORTRANism, but I think
sticking to ARB is ok.


Package description:
The text is perhaps too broken to fix by changing just a bit of the
wording. If this is acceptable to you, I would prefer to take a day to
compose a rewrite that is more to the point and post it for review here.
(And, if you don't mind, I'd then use it upstream as well.)

Elmar