Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 7 Jun 2015 22:36:35 -0400
From: Rich Felker <>
Subject: Re: Build option to disable locale [was: Byte-based C locale,
 draft 1]

On Mon, Jun 08, 2015 at 03:57:01AM +0200, Harald Becker wrote:
> Hi Josiah !
> On 08.06.2015 02:28, Josiah Worcester wrote:
> >To start with: keep in mind that in the case of static linking most of
> >this is not at all pulled in except when strictly necessary. Static
> >linking might be more relevant to your needs.
> You missed my lead in message: I don't talk about static linking
> (which is what I currently do, replacing other stuff). I'm looking
> for a really small shared C library.

Do you have an application in mind where saving ~13k in would
make the difference between being able to use it or not?

> The problem is: Stripping all the functions on every new release is
> too much work, so it needs some minimal support in the library build
> process. Which should be pure optional, enabling the full version as
> default.

Maintaining useless configuration knobs and having to test them all is
also too much work, and the main reason why uClibc is dying.

> >>>- UTF-8 encoding and decoding
> >>May be of use to keep, if on bare minimum.
> >
> >Seeing as the UTF-8 decoder is very small already, I'd be shocked if
> >you could make an argument for removing that.
> That's why I told "keep". I know it to be small, so keep it small
> and fast, and UTF-8 only (not supporting other multi byte char
> sets).

That's how it always has been. You could actually make it smaller by
writing the string functions as trivial character-at-a-time loops on
top of mbtowc/wctomb, but you'd be sacrificing performance to save a
few hundred bytes at most. This does not seem like a worthwhile

> >>>- Character properties
> >>
> >>>- Case mappings
> >>
> >>Keep ASCII, map all none ASCII to a single value.
> >
> >This would be not-quite-right. Also, the case mapping tables are quite
> >small. towctrans.lo which contains the case mappings is 1106 bytes.
> Sorry for my poor English. I mean, let all functions behave as for
> locale "C" (ASCII), but don't fail / break when there is an embedded
> UTF-8 sequence. Just say "this is not ASCII" or may be: "this is
> UTF-8".

musl's "C" locale is _not_ ASCII but is UTF-8. Supporting both
byte-at-a-time operation in the C locale and proper UTF-8 handling
otherwise is the topic of this thread, and it very mildly increases
code size.

> >>>- Internal message translation (nl_langinfo strings, errors, etc.)
> >>
> >>>- Message translation API (gettext)
> >>
> >>No translation at all, keep the English messages (as short as possible).
> >
> >musl does not have any translations in it at all. It only has a small
> >portion of logic able to load external translations. locale_map.lo and
> >__mo_lookup.lo which are together responsible for this, are a total of
> >1471 bytes.
> I mean drop the portion to load external translations.

This probably could be a configurable option at some point, but it's
not going to save you any meaningful amount of space. Best case would
be saving about 4k.

> >>>- Charset conversion (iconv)
> >>
> >>
> >>Copy ASCII / UTF-8, but fail for all other.
> >
> >Though quite possible, it's worth noting that musl iconv is not very
> >large. iconv.lo is 128408 bytes, or 125k.
> A big hunk to kick off. Just keep a stub that allows for ASCII and
> UTF-8 and 1:1 copy operation. Should be possible to draw that down
> to not more than 1k.

119k of the 128k for iconv is legacy CJK character set tables. The
code for them is only 1k or so. Legacy 8bit codepages are another 6k.
The easy way to make things optional here would be just dropping
tables for configured-out charsets.

> >The regex equivalence classes are handled via the isw* functions which
> >(as mentioned above) are quite small.
> So handle them as for the "C" locale, but don't fail/break when
> someone enters an UTF-8 sequence.

"Don't break" means handling UTF-8. Thankfully that's small and
changing this code is not i

> >In short, it seems like if we made these changes we'd maybe be able to
> >trim out 135k and almost all of that would be in iconv. Though I
> >appreciate the desire for smaller code, this doesn't quite seem like
> >the place to go looking.
> My request is, to get a shared lib which does not need to include
> all that other char set and locale code. An optional build option,
> but with the need of some minimal support, so it can persist for
> future releases. I dislike doing all the strip down work over and
> over again for any new release.

Is this a practical need or an ideological one?

> The resulting library shall be standards conform and fully
> operational, as long as the applications only use the bare C locale,
> but should allow to pass through embedded UTF-8 sequences.

Your goal keeps shifting back between supporting UTF-8 and "passing
through sequences" which is _NOT_ supporting UTF-8. If I accidentally
type a non-ASCII character on a command line then press backspace and
the character disappears but there are still two hidden junk bytes on
the command line, that is a broken system. At many levels, you can get
by with just treating text as abstract byte strings that can pass
through anything, but for any kind of visual presentation (even on a
terminal) or entry/editing, you need to know character identity.
Fortunately, as I've been trying to say, that's extremely cheap. You
could save more space by making the string functions or qsort or
something naive and slow...


Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.