Norbert Preining <[EMAIL PROTECTED]> wrote:
> Second: I propose a new question:
> -----------------------------
> Template: tex-common/onlycachefonts
> Type: boolean
> Default: true
> _Description: Apply debconf rules on ls-R files only to /var/cache/fonts/ls-R
> Files under /var/cache/fonts are generated on the fly by all the users
> in case a font is missing. Therefore it is good to allow the update of
> /var/cache/fonts/ls-R by a specified group. If you select yes here, only
> this ls-R file is managed by debconf, all the other get save permissions
> and can only be modified by root. If you choose no, all the ls-R files'
> permission are managed by debconf.
> ----------------------------
Wouldn't it be more appropriate to change the order/hierarchy of
questions a bit? As I understand the wording, the user would only see
this questions after answering "yes" to the general "manage with
debconf" question. I think we should make both questions independent
(and only explain what an "ls-R file" is in the first one). Don't know
which should come first; maybe first about font cache and only ask the
second if they answered yes to the first.
> To my mind only /var/cache/fonts/ls-R should actually have `liberal'
> permissions, while the other ls-R files should not be touchable by
> anybody else the root.
Well, I generally agree, but it's nice to be able to grant ls-R access
to people who also have access to TEXMFLOCAL.
> # move the answers from tetex-bin to tex-common, if there are no answers
> # in the tex-common db
> maybe_move_answer() {
> $ans = $1
> db_get tex-common/$ans || true
> if [ x"$RET" = x ] ; then
> # nothing already saved to tex-common debconf
> db_get tetex-bin/$ans || true
> if [ ! x"$RET" = x ] ; then
> # but there is an answer in tetex debconf
> db_set tex-common/$ans "$RET"
> fi
> fi
> }
Since there are defaults, I think there will always be some return
value. You should rather check for the "seen" flag:
db_fget tetex-bin/$ans seen || true
Also, don't forget to set the "seen" flag to yes for the tex-common
incarnation of the question it it was already seen in tetex.
Regards, Frank
--
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer
