I have previously explained why the attempted destruction of the data by the federal government is unconstitutional (see my posts here, here and here). Having read the written submissions of both parties to the Court of Appeal, I have some further thoughts.
Recall that there are two critical issues in this case. First, did the federal government have the authority to enact s. 29 of the Ending the Long-Gun Registry Act, a provision which requires the destruction of records relating to long guns? If not, second, does Quebec have a right to the records?
The best argument for the federal government is that the Firearms Act constituted a complete code relating to the licensing and registration of firearms. Registration was a function exclusively of a federal official, the Registrar of Firearms. Provincial officers played a role in licensing and other matters, but not in registration. Watertight legislative compartments separated provincial and federal operations. On this approach, the records are the property of the federal government, which is entitled to destroy them along with its repeal of the requirement to register long guns.
Even if this characterization is accurate and not an oversimplification of the regime created by the Firearms Act, there are two serious problems with this argument. First, the Firearms Act itself contained a power to destroy records (ss. 84 and 85). If the Firearms Act indeed constituted a complete code, with watertight compartments that strictly defined provincial and federal operations, then the only power of destruction that the federal government could exercise was the one provided for by the legislation. However, such a power could be exercised only to further the objects of the Firearms Act; it could not be exercised so as to destroy all of the records created. In any event, that provision has now been repealed.
Accordingly -- and this leads to the second serious problem -- the federal government needs to demonstrate an independent jurisdictional basis for its attempt to destroy the data. The argument for the federal government on this point is that destroying the data is part of the repeal of the long-gun registration provisions, carried out with a view to protecting individual privacy and decriminalizing the legitimate use of weapons by hunters.
This argument, however, runs right into Reference re Upper Churchill Water Rights Reversion Act,  1 SCR 297, which appears to be the only Canadian case on point, albeit that it concerned ultra vires action by a province, not by the federal government.
At the heart of the case was a dispute between Newfoundland and Hydro-Quebec about electricity produced at Churchill Falls. Originally, Newfoundland provided by statute for a lease of its resources to a federally-incorporated company. A statutory lease was duly granted. Subsequently, the company entered into a lengthy and lucrative contract with Hydro-Quebec to export its power to Quebec. Later on, though, Newfoundland decided it wanted more power for its own purposes; Hydro-Quebec refused to accede to this request.
Newfoundland's response was to legislate. It repealed the legislation providing for the lease, repealed the statutory lease, and sought to expropriate the assets of the federally-incorporated company.
For present purposes, the most important aspect of the Supreme Court of Canada's unanimous decision is its approval of the following passage from the judgment of the Newfoundland Court of Appeal:
In our view, the Act in question does more than modify or repeal existing legislation. It also purports to expropriate the fixed assets of CFLCo used in the generation of electric power while expressly precluding that company from asserting any claim either for additional compensation for the loss of its property or damages for breach of any of its leases. We must decide, then, whether the legislation is in respect of any of the classes of subjects enumerated in s. 92, and assigned exclusively to the provinces and, if so, whether the subject matter of the Act also falls within one of the classes of subjects in s. 91, as a result of which the legislative authority of the Province is thereby overborne (my emphasis).The Supreme Court went on to hold that the expropriation was ultra vires the province of Newfoundland.
The key point is that the power to repeal legislation does not give the federal government (to which the teachings of Churchill Reversion apply with equal force) carte blanche. Repealing legislation is one thing, rewriting history quite another. Repealing legislation does not require any particular jurisdictional basis. Rewriting history cannot be accomplished without support from a head of legislative competence set out in the Constitution Act, 1867.
Just as Newfoundland could not use its power to repeal legislation to expropriate extant assets, so the federal government cannot destroy extant documents under cover of its (undisputed) power to repeal legislation.
Here, it is difficult to identify a head of power which supports s. 29 of the Ending the Long-Gun Registry Act. The criminal law power used to create the registry cannot support s. 29, because s. 29 does not have the three necessary elements of a criminal law statute: it has no criminal law purpose, it contains no prohibition and it lacks a penalty.
If anything, in its "pith and substance", s. 29 relates to "property and civil rights in the province", a matter within the jurisdiction of the provinces (an argument I laid out in my previous posts).
Finally, there has been some controversy about the first-instance judge's use of the principle of cooperative federalism as a stick with which to beat the federal government. Indeed, using an unwritten constitutional principle as the sole basis for striking down a statutory provision would be inappropriate.
However, the principle of cooperative federalism is useful in interpreting and defining the limits of legislative competence. It underpins, for example, the Churchill Reversion distinction between repealing legislation and rewriting history. It implies that a sweeping power of repeal is inappropriate. And it colours the appropriate interpretation of the Firearms Act, suggesting that there are no watertight compartments, but rather an interlocking and overlapping federal-provincial scheme of mutual assistance and support.
In addition, the principle of cooperative federalism could be useful in resolving the second question in the case: whether Quebec has a right to the data. If s. 29 is unconstitutional, that does not mean the federal government is powerless to do things with the data. But any authority it has must be exercised in accordance with unwritten constitutional principles, including that of cooperative federalism. This is within the conventional understanding of the appropriate sphere of application of unwritten constitutional principles to exercises of discretion (see here, especially paras. 125 and 170-187). It may provide a basis, independent of the first-instance judge's findings of fact on ownership, for compelling the federal government to return the records to Quebec.
* Raluca Popovici has been providing excellent research assistance in my work on this issue. I hope to have a detailed treatment of the issue up in draft form on SSRN at some point over the summer.