Relax checks for extended capability to support new format

[blueyed]

In the 20180331 release, the format was slightly changed:

20180331
+ improve terminfo write/read by modifying the fourth item of the
extended header to denote the number of valid strings in the extended
string table (prompted by a comment in unibilium's sources).

Since the number of valid string capabilities is not necessarily the
same as extstrslen, it's not possible to sanity check the total number
of items up front anymore.

Include a new test with an updated screen terminfo dump.

[blueyed]

Hmm, make test still passes with the new test, but the patch to the source removed?!

diff --git i/unibilium.c w/unibilium.c
index 8586b6f..64472de 100644
--- i/unibilium.c
+++ w/unibilium.c
@@ -339,6 +339,8 @@ unibi_term *unibi_from_mem(const char *p, size_t n) {
             extalllen += extnumlen;
             extalllen += extstrslen;

+            DEL_FAIL_IF(extofflen != extalllen + extstrslen, EINVAL, t);
+
             DEL_FAIL_IF(
                 n <
                 extboollen +

[blueyed]

To be clear: I'd expected the test to fail then.
Since it is not clear to me what the patch is about, at least the test should test for it I think.

[justinmk]

We need this stuff in the 0.4 release, so would be better to merge now and ask questions later ..

ping @jamessan anyway :)

[blueyed]

Yeah :) - but it's not like we need to hurry it "just because", without waiting a few days at least.

[jamessan]

Hmm, make test still passes with the new test, but the patch to the source removed?!

Hmm, guess I didn't choose a good terminfo entry to use for the test.

[blueyed]

@jamessan
So where was the problem seen with then?
Would it happen with newer files from ncurses (development, after 6.1)?
Can we use an existing terminfo entry (from ncurses development) for the test?

[blueyed]

Ok. I've used the screen terminfo from mauke#40, and when generating the test for it it fails without the patch to unibilium.c here:

% TERMINFO_DIRS=$PWD tools/gen-static-test screen > t/static_screen.c
unibi_from_term("screen"): No such file or directory

With this patch it works, and the output is the same.

So the problem with the test not failing is that the patch is needed in the first place to generate the test, which is then using unibi_from_mem directly - but with the result that unibi_dump produced.

Therefore I think the patch is good, but the test adds no benefit (at least with regard to the patch).

[blueyed]

Should be good now.
Without the patch the tests will fail now:

t/new-extended-format-static_screen.t .. 1/541 Bailout called. Further testing stopped: Invalid argument

This ships t/fixtures/s/screen now (from mauke#40).
I guess this might be just an updated version, and we would see the issue with ncurses 6.2 anyway already then - but that's not clear to me.