Fix a bug in gcd for polynomials

[joschmitt]

I don't know what this function is doing, so this is an educated guess. For polynomials over fields, another gcd is used. The function fixed here is only called if one specifies the ignore_content = true.

I found this by calling is_squarefree on a polynomial over a number field in Nemo.

[thofma]

Can you post the example to reproduce it?

[joschmitt]

julia> Rx, x = QQ["x"]
Univariate polynomial ring in x over rationals, (x)

julia> gcd(x, x, true)
ERROR: UndefVarError: `c` not defined
Stacktrace:
 [1] gcd(a::AbstractAlgebra.Generic.Poly{Rational{BigInt}}, b::AbstractAlgebra.Generic.Poly{Rational{BigInt}}, ignore_content::Bool)
   @ AbstractAlgebra ~/.julia/dev/AbstractAlgebra/src/Poly.jl:1890
 [2] top-level scope
   @ REPL[6]:1

gcd(x, x) calls another method though.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.99%. Comparing base (0f40154) to head (f9c6664).
Report is 26 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1799   +/-   ##
=======================================
  Coverage   87.99%   87.99%           
=======================================
  Files         119      119           
  Lines       30073    30073           
=======================================
  Hits        26464    26464           
  Misses       3609     3609           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[thofma]

This version (with the third boolean argument) should probably not be user-facing, but I guess there is no harm fixing it here.

[fingolfin]

could/should we add a test?

[joschmitt]

could/should we add a test?

I don't know what the function is supposed to do, so I don't feel qualified.

[fingolfin]

I don't really follow this whole function, but c is clearly only set if !ignore_content holds, so the existing code can't be right. The new code seems at least plausible.

So let's merge this. If there are problems here, and if anyone is actually using this code, we'll just have to hope for a bug report.