Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.debian.kernel > #61215 > unrolled thread
| Started by | Jonas Meurer <jonas@freesources.org> |
|---|---|
| First post | 2018-06-17 03:00 +0200 |
| Last post | 2018-07-22 22:20 +0200 |
| Articles | 2 — 2 participants |
Back to article view | Back to linux.debian.kernel
Bug#901702: Add locale and gettext support to initramfs Jonas Meurer <jonas@freesources.org> - 2018-06-17 03:00 +0200
Bug#901702: Add locale and gettext support to initramfs Josh Triplett <josh@joshtriplett.org> - 2018-07-22 22:20 +0200
| From | Jonas Meurer <jonas@freesources.org> |
|---|---|
| Date | 2018-06-17 03:00 +0200 |
| Subject | Bug#901702: Add locale and gettext support to initramfs |
| Message-ID | <w1ucW-2Vy-1@gated-at.bofh.it> |
[Multipart message — attachments visible in raw view] — view raw
Package: initramfs-tools Version: 0.130 Severity: normal Tags: l10n patch Hello, when trying to add l10n support to the cryptsetup initramfs script I was surprised to realize that apparently no initramfs component has l10n support yet. Probably that's because most messages from initramfs are more targeted at developers. Unfortunately, that's not true for the initramfs scripts of cryptsetup. Here, we need to ask users for input (passphrase) and therefore the messages we print are targeted at normal users. There's also an open bugreport that requests support for translated strings[1]. I first thought about adding all the required locale and gettext stuff to initramfs in the cryptroot hook, but after thinking about it again, I think it should be done in initramfs itself instead. Other initramfs components might want to use it as well. That's why I'd like to add locale and gettext support to initramfs but make it optional. Without any prompts targeted at endusers, there's no real need to bloat the initramfs with locales and gettext files. But e.g. in the cryptsetup package, we would enable it. I've prepared a patch that optionally adds locale and gettext support to initramfs (depending on a initramfs.conf variable). You can find the patch attached to this bugreport or as a merge request on Salsa[2]. Whatever you prefer ;) Would be awesome if you could consider to merge it. It's a prerequisite for adding l10n support to the cryptsetup initramfs scripts. Cheers, jonas [1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=688735 [2] https://salsa.debian.org/kernel-team/initramfs-tools/merge_requests/4
[toc] | [next] | [standalone]
| From | Josh Triplett <josh@joshtriplett.org> |
|---|---|
| Date | 2018-07-22 22:20 +0200 |
| Message-ID | <wesZH-6iy-1@gated-at.bofh.it> |
| In reply to | #61215 |
On Sun, 17 Jun 2018 02:53:15 +0200 Jonas Meurer <jonas@freesources.org> wrote: > That's why I'd like to add locale and gettext support to initramfs but > make it optional. Without any prompts targeted at endusers, there's no > real need to bloat the initramfs with locales and gettext files. But > e.g. in the cryptsetup package, we would enable it. > > I've prepared a patch that optionally adds locale and gettext support to > initramfs (depending on a initramfs.conf variable). You can find the > patch attached to this bugreport or as a merge request on Salsa[2]. Reading through the patch, most of it seems reasonable; however, I noticed a few issues: - /usr/bin/locale isn't in the locales package, it's in libc-bin, which is essential. So, please don't point people towards the locales package. - You're filtering out LC_COLLATE from the locale output, but that just leads LC_COLLATE to default to the same locale specified by other variables. If you want to avoid referencing the collation files, you need to *explicitly* set LC_COLLATE=C. - For the purposes of localized message support *only*, could you do without the (large) LC_CTYPE file as well? Or perhaps more generally, would it make sense to *only* support LC_MESSAGES and nothing else? - For invoking find, please see how other hook scripts do so, and please quote the resulting filenames. Yes, files under /usr/lib/locale should never contain whitespace, but despite that, please don't encode such assumptions. - Nit: rather than providing a function gettext_support to conditionally source /usr/bin/gettext.sh and otherwise provide dummy functions, have you considered just putting a /usr/bin/gettext.sh in the initramfs that defines those dummy functions, so that scripts can use it unconditionally? Have you confirmed that this support works on a system with locales not installed and only the C.UTF-8 locale configured?
[toc] | [prev] | [standalone]
Back to top | Article view | linux.debian.kernel
csiph-web