#pragma GCC optimize(2)
#include<bits/stdc++.h>
using namespace std;
typedef long long ll;
typedef __int128_t int128;
typedef pair<int, int> pii;
#define iofast ios::sync_with_stdio(false),cin.tie(0),cout.tie(0)
#define lowbit(x) (x&(-x))
#define inv(x) qpow(x,mod-2)
#define debug(x) cout << (#x) << " = " << x << endl
#define range(x) (x).begin(), (x).end()

const int iINF = 0x3f3f3f3f;
const ll llINF = 0x3f3f3f3f3f3f3f3f;
const ll mod = 998244353;

/*---------------------------------------*/

void solve()
{
    int n, v;
    cin >> n >> v;

    vector<pair<ll, ll>>p[2];
    for (int i = 1;i <= n;i++) {
        ll a, b;
        cin >> a >> b;
        if (a > 0)
            p[0].emplace_back(a, b);
        else
            p[1].emplace_back(a, b);
    }
    // cout << p[0].size() << '\n';
    long double l = 0, r = 1e13;
    for (int cnt = 1;cnt <= 1000;++cnt) {
        long double mid = (l + r) / 2;
        long double pl = 1e28, pr = -1e28;
        for (const auto& [x, y] : p[0])
            if (pl - x + mid * y >= -1E-18L)
                pl = x - mid * y;
        for (const auto& [x, y] : p[1])
            if (x + mid * y - pr >= -1E-18L)
                pr = x + mid * y;

        if (pr - pl >= -1E-18L)
            r = mid;
        else
            l = mid;
    }

    // printf("%.15LF\n", v * l) << '\n';
    // cout << p[0][0].first - l * p[0][0].second << '\n';
    cout << fixed << setprecision(15) << v * l << '\n';
    // cout << setprecision(15) << (1.230769230769231L - 1.23076923L) / 1.23076923L - 1e-8L << '\n';

}

/*---------------------------------------*/

int main()
{
    iofast;
    int t = 1;
    cin >> t;
    // cout << "38.25\n1.23076923\n3000000000000";
    while (t--)
        solve();

    return 0;
}