559706 - Use of <algorithm> in jsworkers.cpp breaks android build

Status: RESOLVED WONTFIX

(Core :: JavaScript Engine, defect)

Description

[Michael Wu [:mwu]]

Bionic doesn't provide algorithm.

Comment 1

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

We'll need a workaround here for merging, even before we get stlport or something similar in the tree.

Comment 2

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Oh right, this is just in shell workers... we can probably just disable shell workers on Android temporarily.

Comment 3

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Attachment: disable shell workers for android for now

We can disable this for android temporarily, until we get a version of STL in-tree.

Comment 4

[Jason Orendorff [:jorendorff]]

Comment on attachment 445469 [details] [diff] [review]
disable shell workers for android for now

Well ... all right, but I think I'd rather see jsworkers.cpp grow a portable implementation of reverse() and stop including <algorithm>. It might go like this:

template <class T>
void swap(T &a, T &b)
{
    T tmp = a;
    a = b;
    b = tmp;
}

template <class It>
void reverse(It begin, It end)
{
    while (begin != end) {
        swap(*begin, *end);
        if (++begin == end)
            break;
        --end;
    }
}

Comment 5

[Jason Orendorff [:jorendorff]]

mrbkap points out that this reverse() implementation has an off-by-one error: it needs to decrement end before the first swap.

Comment 6

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Attachment: better

Ok, this implements a local reverse() impl.  Tested this by running jstests.py with -j 20 -- is that sufficient?  Wasn't sure how to exercise the implementation.

Comment 7

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Hm, or maybe that's not enough, because -j doesn't seem to be jsworkers but separate instances of js.exe (the --help docs say "WORKER_COUNT"!).  How can I test this?

Comment 8

[Jason Orendorff [:jorendorff]]

Comment on attachment 448140 [details] [diff] [review]
better

Thanks.

Comment 9

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

wontfixing this -- we'll have workarounds for STL in android (initially an alternate ndk, eventually an in-tree STL).