Chapter 5 binutils nitpick

[randy@…]

In the Note in section 5.4.1 it says:

"To achieve this easily, wrap the four commands in a time command like this:"

However, there are only 3 commands shown: configure, make and make install.

[randy@…]

Another nit about this same area. It says under --disable-nls:

"This disables internationalization. This is not needed for the temporary tools."

This could be confusing to non-native English speakers. It very well could be interpreted as "This disables i18n, this also is not needed for the temporary tools". Which means folks might be confused as to why it is ever there.

Suggested new wording:

"This disables internationalization as i18n is not needed for the temporary tools."

[archaic@…]

Fixed in trunk r5996 and testing r5997.

[archaic@…]

Forgot about the initial problem and only fixed comment #1. Reopening.

[Matthew Burgess]

Hmm, maybe if folks didn't post directly to the list instead of commenting in bugzilla I'd be able to track the suggestions a little easier :)

Archaic, your suggestion in ​http://www.linuxfromscratch.org/pipermail/lfs-book/2005-June/014952.html of:

"time { ./configure ... && ... && ... && make -C ld LIB_PATH=/tools/lib; }"

makes sense to me. I count 5 commands there: "../binutils-$version/configure && make && make install && make -C ld clean && make -C ld LIB_PATH=/tools/lib. So it looks like we need to s/four/five/ and add an extra set of '&& ...' to your suggestion above.

[archaic@…]

I don't know that being pedantic is really necessary what with the ...'s, but I'll fix this shortly.

[randy@…]

Is there any way we can keep the binutils calculation the same here in the Testing branch?

By adding more commands to the initial binutils calculation, though it isn't that much time (it took another 30 seconds to do make ld clean and make ld on a slow machine), this skews *all* the SBUs in LFS and BLFS.

Could we implement Matt's suggestions in trunk and just change the 4 to a 3 in testing so the SBU calculations remain valid?

[Matthew Burgess]

Hmmm, I stupidly forgot this was where we derive all the SBUs from. Let's just s/four/three/ and change the `time' description to:

'./configure ... && ... && make install'. I think that then makes it clearer that the 2 commands at the bottom of the page are additional to the usual compilation/installation procedures. Comments?

[archaic@…]

I can't say I agree. Everything in grey boxes should be construed as the "build time" of binutils and indeed is how I have been calculating SBU's since the beginning when it was bash not binutils. Patching, sedding, everything except untarring and later removing the sources. Otherwise, what possible standard could a reader expect? One possible exception (and this can be standardized) is testsuites. Section 4.5 can state everything but testsuites is included. Either way, as it is there is no standard and since package instructions can vary wildly, using all the instructions on the page (sans testing) is the only standard I see that is maintainable.

However, for the sake of 6.1, we could drop the wording for 4 to 3 since that is how Randy has calculated the current BLFS svn.