Conversion from formal series to lazy power/LaurentSeriesRing give confusing error message

[user202729]

Steps To Reproduce

sage: var("x")
....: f = (1/x+sqrt(x+1)).series(x)
x
sage: f
1*x^(-1) + 1 + 1/2*x + (-1/8)*x^2 + 1/16*x^3 + (-5/128)*x^4 + 7/256*x^5 + (-21/1024)*x^6 + 33/2048*x^7 + (-429/32768)*x^8 + 715/65536*x^9 + (-2431/262144)*x^10 + 4199/524288*x^11 + (-29393/4194304)*x^12 + 52003/8388608*x^13 + (-185725/33554432)*x^14 + 334305/67108864*x^15 + (-9694845/2147483648)*x^16 + 17678835/4294967296*x^17 + (-64822395/17179869184)*x^18 + 119409675/34359738368*x^19 + Order(x^20)


sage: LazyPowerSeriesRing(QQ, "x")(f)
<repr(<sage.rings.lazy_series_ring.LazyPowerSeriesRing_with_category.element_class at 0x712d8ec95f40>) failed: TypeError: Substitution using function-call syntax and unnamed arguments has been removed. You can use named arguments instead, like EXPR(x=..., y=...)>
sage: LazyLaurentSeriesRing(QQ, "x")(f)
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
...
ValueError: the valuation must be specified

Related but different: #39839, #39809

Expected Behavior

The conversion should redirect the user to use PowerSeriesRing or LaurentSeriesRing instead

Environment

• OS: Linux
• Sage Version: latest

Checklist

• I have searched the existing issues for a bug report that matches the one I want to file, without success.
• I have read the documentation and troubleshoot guide

[tscrim]

This actually doesn't make sense because the input is not exact; it truncates the data to a finite amount. As such, @mantepse should we forbid it and similar such things (e.g., finite precision power series) as input?

[mantepse]

I agree with @tscrim. I encountered various issues with the element constructors in the lazy series framework, I think it would be good to make them more uniform and possibly a bit stricter.

This term I won't be able to work on it, though.

[user202729]

What do you mean it doesn't make sense? Elements of LazyPowerSeriesRing can have finite precision too.

sage: LazyPowerSeriesRing(QQ, "x").zero().add_bigoh(3)
O(x^3)
sage: LazyPowerSeriesRing(QQ, "x").zero().add_bigoh(3)[5]
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
Cell In[4], line 1
----> 1 LazyPowerSeriesRing(QQ, "x").zero().add_bigoh(Integer(3))[Integer(5)]

File power_series_poly.pyx:463, in sage.rings.power_series_poly.PowerSeries_poly.__getitem__()

IndexError: coefficient not known

(Granted you can ask why doesn't the user not just use PowerSeriesRing instead? Still it doesn't seem too far-fetched to support this, one possible use case is because some other lazy series is used elsewhere)

[tscrim]

No, they have infinite precision. You should look at the type/parent after the add_bigoh() call.

[tscrim]

Or to put it another way, if you do not call add_bigoh() (which would keep the power series computed lazily), then you would not get an error.

[user202729]

You're right here. I suppose we can improve the error message to inform the user to use normal PowerSeriesRing instead.