dissenting:

Rolando Ruiz’s state habeas counsel was found to be “wholly incompetent” and “egregiously inept,” and I do not doubt that he was. Whether ineffective habeas counsel excuses Ruiz’s failure to exhaust his ineffective trial counsel claim before Texas’s courts, so as to prevent the claim from being procedurally barred here, is a difficult question. It was a difficult question when the district court answered it in the negative in 2005, when this panel affirmed the district court’s judgment in 2006, and when it was presented to the Texas Court of Criminal Appeals (“TCCA”) eight days before Ruiz’s scheduled execution this year.

But nothing has changed since the district court rendered its opinion that warrants the extraordinary measure of setting it aside. At the very least, not enough has changed to make the district court’s refusal to set it aside an abuse of discretion. The majority believes that the TCCA’s ambiguous one-page per curiam order, which it had all of four days to render given the impending execution, compels Rule 60(b) relief. I disagree. I also believe the fact that Ruiz’s first presentation of this claim to the Texas courts came just eight days before his scheduled execution supports the district court’s finding that he was not diligent. I respectfully dissent.

I. DISCUSSION

A. Rule 60(b) and the TCCA’s Opinion

Rule 60(b) provides six grounds for relief from a prior judgment. While Ruiz does not specify which ground he is pro*533ceeding on, the only relevant ground is the catchall provision allowing for relief from a judgment based on “any other reason justifying relief from the operation of the judgment.” Fed. R. Civ. P. 60(b)(6). Such relief is granted rarely and only under extraordinary circumstances. Batts v. Tow-Motor Forklift Co., 66 F.3d 743, 745 (5th Cir. 1995).

There is no dispute that the district court’s first judgment denying relief on procedural default grounds was correct when it was issued in 2005. Ruiz’s counsel at oral argument agreed that it was a reasonable decision, and this court has already affirmed it. But Ruiz points to the TCCA’s most recent order denying relief in this case as an intervening factor that now renders that judgment inequitable. His argument is that the latest TCCA opinion was ambiguous, which compels the district court to treat it as merits based under the directives of Long and Coleman, therefore requiring the district court to entertain the merits of the habeas petition. Michigan v. Long, 463 U.S. 1032, 103 S.Ct. 3469, 77 L.Ed.2d 1201 (1983); Coleman v. Thompson, 501 U.S. 722, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991).

While the majority finds that argument convincing, I do not believe that Long&nd Coleman apply to this case. I agree with the majority that the TCCA’s “decisional basis here is uncertain,” and if we were reviewing that decision we would be required to treat it as merits based and, accordingly, reach the merits ourselves. But that is not the posture this case comes to us in. Ruiz has already received federal habeas review, and now brings this as a Rule 60(b) motion for relief from the earlier judgment.

The question on a Rule 60(b) motion for relief is much different than in the Coleman context. A Coleman petitioner typically brings his habeas claims to the state courts first. If relief is denied, then the district court may entertain a federal ha-beas claim unless the state court’s judgment was based on adequate and independent state grounds. Coleman, 501 U.S. at 730-32, 111 S.Ct. 2546. The Coleman presumption comes into play when evaluating whether the state court’s previous opinion was based on independent state grounds; it instructs courts to treat any ambiguous state decisional basis as a ruling on the merits, thereby permitting federal review. Id. at 734-35, 111 S.Ct. 2546.

Here, Ruiz first brought his unexhaust-ed ineffective assistance claim to federal court. Because there was no state court judgment to interpret through the Coleman presumption, the district court had to answer a hypothetical question: “Would the state court to which the prisoner would have to present his claims in order to exhaust them find the claims procedurally barred?” Kittelson v. Dretke, 426 F.3d 306, 315 (5th Cir. 2005). The district court answered yes, and we agreed.

The relevant question on this Rule 60(b) motion is not whether the TCCA’s subsequent opinion was unequivocally based on independent state-law grounds, as it is in the Coleman context and under the majority’s treatment. The question is whether the TCCA’s one-page opinion creates the “extraordinary circumstances” required to set aside the district court’s otherwise unassailable judgment.1 See Batts, 66 F.3d at 745. This is an extremely difficult standard to meet. For example, it is clear that even “a change in state decisional law, *534rendered after this court makes an Eñe prediction, will not normally constitute an extraordinary circumstance, and cannot alone be grounds for relief from a final judgment pursuant to Rule 60(b)(6).” Id. at 748.

Even if a definitive change in law did constitute grounds for relief, Ruiz cannot make that showing. He never demonstrates a change in Texas law that clearly undermines the validity of the district court’s judgment. Perhaps the TCCA’s opinions in Ex Parte Hood and Ex Parte Campbell show that Ruiz’s argument was not entirely futile before the state courts— as he assumed it to be — but that hardly merits setting aside the district court’s prior judgment when it still appears to be correct, or at least reasonable. Ex Parte Hood, 211 S.W.3d 767 (Tex.Crim.App. 2007); Ex Parte Campbell, 226 S.W.3d 418 (Tex.Crim.App. 2007).

Ruiz focuses on an unclear one-page opinion that was rendered after only four days of consideration due to Ruiz’s fast-approaching execution.2 In my view, that does not cast sufficient doubt on the district court’s judgment to warrant this exceptional relief, much less to find that the district court abused its discretion by not granting relief. While ambiguity in the TCCA’s judgment compels reaching the merits in the Coleman context, ambiguity in this context, where it is incumbent on Ruiz to show exceptional circumstances warranting relief, cuts the other way. Ruiz brought his unexhausted claim to the district court, forcing it to make an Eñe guess on the procedural default issue. It decided against him. Since then, nothing has occurred to demonstrate that the court’s Eñe guess was actually wrong. An ambiguous TCCA opinion under extreme time constraints is not the type of extraordinary circumstance compelling Rule 60(b) relief.

B. Diligence and Balancing the Equities

Rule 60(b)(6) “is a grand reservoir of equitable power,” allowing for consideration of all factors that speak to equity. Harrell v. DCS Equip. Leasing Corp., 951 F.2d 1453, 1458 (5th Cir. 1992). This leaves the district court with extensive discretion to weigh equitable considerations, and I cannot agree with my colleagues that it abused its discretion when it found Ruiz was not diligent enough in pursuing his unexhausted claim before Texas’s courts to warrant relief.

Ruiz knew he had an unexhausted ineffective assistance claim in February, 2004, when he proceeded to federal court. He should have exhausted his claim in the Texas courts in a second state habeas petition. If his petition was denied as an abuse of the writ, he then could have pressed his argument before the federal courts that he had a valid legal excuse for not raising his ineffective assistance claim in his first state petition. Notably, AED-PA’s statute of limitations would have been tolled during any second state habeas proceeding. Villegas v. Johnson, 184 F.3d 467, 468 (5th Cir. 1999). The only explanation Ruiz offers for not taking this ap*535proach is that it was futile to press the ineffective assistance claim in a second ha-beas petition because it would have been dismissed as an abuse of the writ. However, this argument is at least slightly at odds with his first one. See supra Part I.A. Ruiz’s explanation for not exhausting his claim in a second state petition is that it would have been futile, which makes it exceptionally odd, perhaps inequitable, to now hear him complain that the district court was wrong to reject a claim that he considered frivolous. He does point to some intervening state caselaw to explain this anomaly, but none of it seems significant enough to take his claim from being completely frivolous to being correct.

That Ruiz waited more than three months after certiorari was denied in his federal case to file his second state petition also works against him. While three and a half months is not normally an unreasonable amount of time for such a filing, Ruiz had already gone through several proceedings regarding this exact ineffective assistance claim. Waiting until eight days before his scheduled execution is somewhat concerning, especially when he now seeks to benefit from the resulting imprecision in the TCCA’s rushed order.

I do not think it is easy to balance all of the equities in this case. I might disagree with how the district court balanced them, but I do not think it stepped outside the bounds of its discretion here when it found that Ruiz was not diligent and that the equities did not compel relief from judgment.

II. CONCLUSION

I would affirm the district court’s order denying Ruiz’s Rule 60(b) motion and deny any further stay of his execution.

This is a death penalty case from Bexar County, Texas. Petitioner Rolando Ruiz appeals the federal district court’s order denying his Rule 60 motion for relief from judgment and stay of execution. Ruiz brings serious charges of incompetent and ineffective trial counsel, but no federal court has considered the merits of his constitutional claims, and he obtained a stay only within minutes of his execution, granted by this court in order to gain sufficient time to consider properly the appeal. After further briefing and oral argument, we continue the stay of execution, reverse the judgment dismissing the federal habeas petition and remand with instruction to the federal district court to decide the claim of ineffective trial counsel on its merits after any further proceedings necessary to do so.

I

Rolando Ruiz was convicted of murder by a Texas jury and sentenced to death. At the state’s urging, and over his objections, the state trial court declined to appoint different counsel on direct appeal on the promise that any Sixth Amendment claim of ineffective counsel could be raised by state habeas counsel. It was not. Without meeting his client or conducting any investigation, Ruiz’s state habeas counsel filed a boilerplate application in state court that did not challenge the failure of his trial counsel to investigate and *525present his mitigating evidence. The Texas Court of Criminal Appeals (CCA) denied relief.

Ruiz turned to federal court and with a new lawyer filed a federal habeas application raising claims of ineffective assistance by trial counsel and his state habeas counsel. The district court denied relief due to “[t]he inexplicable failure of petitioner’s state habeas counsel to raise any of these claims during petitioner’s state habeas corpus proceeding.” The district court ruled that the claim was procedurally defaulted because “the state court to which he would be required to petition would now find the claims procedurally barred.” Although convinced that while Ruiz’s state counsel was “wholly incompetent” and “egregiously inept,” the district court rejected Ruiz’s contention that this inept performance gave legal excuse for the failure to exhaust. It refused to entertain the claim of ineffective assistance of trial counsel and refused to issue a certificate of appealability for that claim. The federal district court judge did issue a certificate of ap-pealability upon a claimed error in juror selection. Then, in denying Ruiz’s motion to alter judgment, the district court made clear that his ruling rested on his conclusion that “Texas law precludes petitioner from obtaining a ruling on the merits of his currently unexhausted claims ... in a successive state habeas corpus application.” Ruiz’s federal counsel asked the federal district court to stay the federal proceeding to allow Ruiz to return to state court to exhaust the ineffective-assistance claim, pointing to the CCA’s then-recent abandonment of its refusal to accept a state court habeas application so long as the petitioner had a federal habeas petition pending. In refusing to hold the federal claim in abeyance, the federal district court observed:

[E]ven if this court were to hold this cause in abeyance so as to permit petitioner to fairly present the state habeas court with his unexhausted claim herein for the first time, such an action would be an exercise in futility .... Petitioner’s reference to the recent recision of the long-standing Texas rule which prohibited simultaneous litigation of a claim in both state and federal habeas corpus proceedings is non sequitur.1

Armed with the COA on the juror issue, Ruiz appealed to this court, at the same time seeking review of the district court’s rejection of legal justification for the failure to exhaust. We affirmed and the Supreme Court denied certiorari.

With the failure to persuade the federal courts that his lack of effective habeas counsel excused his failure to exhaust, and with suggestions by the CCA that a return to state court might no longer be futile, approximately ninety days later Ruiz filed an application for state habeas relief raising his claim of ineffective assistance at trial and in the presentation of his first state habeas application. The CCA dismissed his application with an order on which much now depends, as we will explain. After the CCA’s dismissal of his application, Ruiz returned to federal district court, filing a motion under Rule 60(b) for relief from judgment in light of the CCA’s ruling, which, Ruiz argued, rejected on the merits his Sixth Amendment claim of ineffective trial counsel and in doing so pulled the ground from under the federal district court’s earlier judgment dismissing the claim and refusing to hold the federal claim in abeyance while Ruiz returned to state court with his unexhaust-ed claim. The district court defended its earlier dismissal and denied the motion, rejecting the contention that the CCA’s *526decision had not rested upon an independent and adequate state law ground. As we will explain the latter was an error of law and the former an abuse of discretion.

II

We first address whether Ruiz’s Rule 60(b) motion is subject to the additional restrictions that apply to “second or successive” habeas corpus petitions under AEDPA.2 We are not persuaded that the motion was a successive petition.

Rule 60(b) allows a losing party to seek relief from judgment under a limited set of circumstances including fraud, mistake, and newly discovered evidence.3 Relief is available under Rule 60(b) in habeas proceedings, but of course only in conformity with AEDPA, including its limits of successive federal petitions.4 It is a subsequent habeas corpus application whenever the Rule 60 motion presents a “claim” for ha-beas relief. The Supreme Court has provided guidance, holding that “[i]f neither the motion itself nor the federal judgment from which it seeks relief substantively addresses federal grounds for setting aside the movant’s state conviction, allowing the motion to proceed as denominated creates no inconsistency with the habeas statute or rules.”5 The Court provided helpful examples, explaining that a Rule 60(b) motion is a habeas claim when it presents a new claim for relief, or when it presents new evidence in support of a claim already litigated,6 or when it asserts a change in the substantive law governing the claim, or when it attacks the federal court’s previous resolution of a claim on the merits. Significantly, the Court then explained that there is no new habeas claim “when [a petitioner] merely asserts that a previous ruling which precluded a merits determination was in error — for example, a denial for such reasons as failure to exhaust, procedural default, or statute-of-limitations bar.”7

Such is the case here. The federal district court’s previous denial of Ruiz’s claim was not “on the merits.” That is, the district court did not rule that there were no grounds entitling Ruiz to habeas corpus relief under 28 U.S.C. §§ 2254(a) and (d), but rather denied relief based on procedural default and failure to exhaust, two rulings specifically identified by the Court as rulings precluding a merits determination. So the district court had jurisdiction to consider Ruiz’s Rule 60(b) motion, free of the jurisdictional constraints of AEDPA upon successive petitions. In short, Ruiz is pursuing his first federal petition with its claim that his trial counsel was ineffective in failing to investigate and otherwise develop a mitigation case,8 a “Wiggins” claim.

*527III

The federal court below would have had no federal claim before it if, as the district court concluded, the CCA applied an independent and adequate state law ground to deny relief.9 Moreover the petition would then have been successive and Rule 60 would not allow Ruiz to escape the successive writ limits of AEDPA.

In deciding whether the CCA refused relief upon an independent state-law ground or upon the merits of Ruiz’s petition we áre aided by the bright light of Michigan v. Long:

when ... a state court decision fairly appears to rest primarily on federal law, or to be interwoven with federal law, and when the adequacy and independence of any possible state law ground is not clear from the face of the opinion, we will accept as the most reasonable explanation that the state court decided the case the way it did because it believed that federal law required it to do so.10

This settled principle gives to state courts control over the federal review of their opinions. It has become a rote rule at the fingertips of every writing member of state courts of last resort — where studied ambiguity or clarity in the decisional footing is an art form and an absence of clarity in an opinion is seldom inadvertent. Calibrated uncertainty can play a mediating role in garnering support for an outcome. To the point, that the CCA did not make clear that its decision rested on an independent state ground opens the merits of Ruiz’s Wiggins claim to federal review. At best, the CCA did not make clear whether it relied on state or federal law in dismissing Ruiz’s application. As the CCA is keenly aware, its choice of language was made against a background legal standard — which directs the CCA in either granting an application for consideration of subsequent claims or dismissing that application as an abuse of the writ — that is interwoven with federal law.

The Texas Code of Criminal Procedure, as interpreted by the CCA, provides for subsequent applications where (1) the factual or legal basis for the subsequent claim was previously unavailable and (2) where the facts alleged would constitute a federal constitutional violation that would likely require relief from either the conviction or sentence.11 The boilerplate dismissal by the CCA of an application for abuse of the writ is itself uncertain on this point, being unclear whether the CCA decision was based on the first element, a state-law question, or on the second element, a question of federal constitutional law.

In any event the decisional basis here is uncertain. Although only seven members of the court participated, joinder of five judges remained necessary for a decision.12 So even assuming that the four-judge plurality rejected Ruiz’s application as an abuse of the writ based on the first element, there is still no fifth vote for an independent state-law ground. Judge Womack provided the fifth vote for dismissal, filing a concurring opinion in order *528to reserve the question of whether Ruiz’s procedural default should be excused by the unreasonable failure of state habeas counsel. His concurring opinion explicitly rests on the conclusion that Ruiz did not allege a meritorious Sixth Amendment claim. He saw the claim of ineffective trial counsel to be a failure of presentation, not a failure to investigate, the latter according trial counsel far less deference than that given to tactical decisions of presentation. Even if it were clear, the four-Judge plurality was not controlling.

Setting aside the intricacies of the CCA’s voting requirements, both the concurring and dissenting opinions, by their unanswered language, strongly suggest that the CCA debated and reached the federal merits question, not the independent state law ground. The two dissenters thought Ruiz was presenting a Wiggins claim (failure to investigate), chiding the plurality of four for failing to understand that Ruiz presented this breed of ineffective-assistance claim, more easily sustained than a claim of poor choices in presenting mitigating evidence.

The district court concluded that Judge Womack’s concurring opinion is “utterly non sequitur” unless the plurality is read to rest on state-law grounds. We are not persuaded. The concurring opinion can also be read as relaying the decision path of the court, so as to preserve for another day the question whether the obvious deficiencies in the performance of state habeas counsel excused a failure to present the claims of ineffective trial counsel in a first state petition. Preserving this issue makes sense only if the CCA had rejected the claim of ineffective trial counsel on its merits rather than because it was not presented in Ruiz’s first state habeas petition. In sum, the three opinions from the seven judges together do not clearly rest on an independent and adequate state ground. Even if the order of the four-Judge plurality alone left the decisional footing certain, and it did not, Judge Womack’s opinion, necessary to the court’s judgment, pushes the court toward a clear merit ruling, and in any event deprives the plurality of a fifth vote on an independent and adequate state ground. This leaves the decisional path far short of the clarity insisted upon by Michigan v. Long, to which the district court paid no mind and of which the CCA is acutely aware. Ruiz’s Wiggins claim was properly before the federal district court.

IV

The district court also concluded, in a final paragraph, that the balance of equities tips away from the defendant. We do not agree.

Ruiz’s federal habeas petition claimed that he should be excused under AEDPA13 for failure to exhaust and that he satisfied the “cause” exception to procedural default. He explained that Texas affirmatively interfered with his efforts to obtain new counsel for direct appeal and obstructed his efforts to prosecute his ineffective-assistance claims by appointing incompetent state habeas counsel. On August 29, 2005, the federal district court rejected these arguments, denying Ruiz’s ineffective-assistance claim as unexhausted and procedurally defaulted, while granting COA on a “Witherspoon” point.14

As we have recounted, Ruiz pressed his contention that the district court erred in failing to find legal excuse for his failure to exhaust the claim of ineffective trial counsel, appealing the denial of COA to this *529court and then seeking certiorari. The Supreme Court denied certiorari on March 19, 2007. Approximately ninety days after certiorari was denied, Ruiz returned to state court with his ineffective assistance claims. The CCA dismissed the application on July 6, 2007, with the order we have discussed, and Ruiz asked the federal district court to reconsider its earlier order dismissing for failure to exhaust.

After rejecting the contention that the CCA had rejected his constitutional claims on their merits, which meant that Rule 60 offered no relief, the district court chided Ruiz for a perceived lack of due diligence in not immediately presenting his unex-hausted ineffective-assistance claims to the state court when it earlier dismissed them. Pointing to his perceived lack of due diligence, the district court insisted that Ruiz should have returned to state court despite its judgment that state-court litigation would have been futile given the CCA’s summary dismissal as abuse of the writ.

While the district court’s conclusion of futility was sound when made, it has been undermined by recent decisions by the CCA. In January 2007, the CCA decided Ex parte Hood, indicating for the first time that there are judicially-created exceptions to section five.15 Before this decision, neither Ruiz nor the district court had reason to believe that the CCA would create an equitable exception to the successor bar. On April 25, 2007, the CCA decided Ex parte Campbell, which, as we explained above, held that the Texas procedural bar based on factual unavailability incorporates a question of federal constitutional law.16 Before this decision, neither Ruiz nor the district court had any basis to view the state successor provision as anything but an independent and adequate state ground. Ruiz’s attorney explains, persuasively, that without these recent decisions he would not have donated his time to this case.

Ironically it was Ruiz who brought the modification of Texas’s two-forum rule to the attention of the district court and asked that it hold the federal case to allow him to exhaust.17 As we explained, the district court refused to hold the case, observing that a return to state court would be futile. Rather, it granted COA on another point of law.

We cannot agree that Ruiz was dilatory in pursuing the appeal of the district court’s rejection of his proffered excuses for default under federal law as well as the Witherspoon claim for which the district *530court had granted COA. Ruiz chose the only avenue available to him, as the district court correctly observed, and litigated in federal court his obligation to exhaust. The district court also correctly observed that Texas modified its two-forum rule that forbade filing for relief while federal relief was also being sought, however Ruiz was nonetheless left to pursue a mixed petition or to abandon the constitutional claim on which the district court had granted COA. Texas only opened its courts to petitioners with pending federal claims when those claims had been stayed by the federal courts. The district court, focused on the futility of Ruiz state litigation, refused the required stay, closing the door that it now concludes Ruiz should have walked through. In practical terms the district court’s view of due diligence insisted that Ruiz should have abandoned his federal challenge to a duty to exhaust the claim of ineffective trial counsel. We do not understand due diligence to require a litigant, while diligently pursuing a final federal ruling that he was excused from exhausting his claim, to nevertheless exhaust the claim in state court — emphatically so when he was unable to obtain the requisite stay of federal proceedings and therefore would have had to abandon his effort to claim that his failure to exhaust should be excused, all a futile act.

A court may consider the last-minute nature of an application to stay execution in deciding whether to grant equitable relief,18 but we do not see this as a last-minute attempt to manipulate the judicial process. The district court, overlooking its closure of the door to state court by refusing a stay, failed to consider that Ruiz’s return to state court followed on the heels of the earliest suggestion that doing so would not be futile. Unlike Harris, Ruiz’s petition does not come to us after sitting on his rights for eighteen years.19 The state of Texas interfered with Ruiz’s rights until March 30, 2003, when his state habeas counsel was finally terminated and replaced with counsel from Texas Defenders. These lawyers diligently pursued Ruiz’s claims in federal court until certio-rari was denied three months before Ruiz’s execution date. The state of Texas interfered with Ruiz’s constitutional rights at every critical turn in this litigation, appointing ineffective trial counsel who failed to investigate his past; re-appointing the same on direct appeal, despite Ruiz’s pleas for a new lawyer; and finally providing Ruiz with incompetent state habeas counsel, who presented the CCA with “a set of boilerplate, frivolous claims.”20 Indeed the federal district court said it best, characterizing Ruiz’s underlying ineffective-assistance claim as “significant, potentially meritorious”21 and calling his state habeas counsel “appallingly inept” and “egregiously deficient.”22 We are given no rational reason to conclude that the equities run against Ruiz, despite the “fundamental unfairness” of his habeas proceedings, or why the district court now faults the lawyers who first exposed the work of previ*531ous counsel as being dilatory. Whatever be the explanation, in this difficult area of the law of capital punishment, we are met with the inescapable conclusion that the district court’s balancing of equities was infected by its first holdings — that the claims of ineffective assistance of trial counsel were again not before it and that it had been correct in holding that the state court would not decide the case on a return to it. And of course if this first conclusion of the district court were sound then it lacked jurisdiction over the resulting “successive writ.”

Texas argues that in balancing the equities we are bound to ignore the performance of Ruiz’s state-habeas lawyer. As Texas notes, it is well settled that “ineffective state habeas counsel does not excuse failure to raise claims in state habeas proceedings.”23 From this rule, Texas induces that “Ruiz cannot now claim that state habeas counsel’s conduct somehow excuses his failure to diligently pursue relief in state courts, when that same conduct has been found not to constitute cause for Ruiz’s failure to exhaust the claims to the state courts first.” Texas also argues that in balancing the equities we are bound to ignore the earlier futility of state court litigation. As Texas notes, it is well settled that futility does not excuse a litigant’s failure to assert- a claim in state court unless the very basis of that claim was unavailable.24

Both these arguments are axiomatic, but answer the wrong question. We are not dealing with the equities of considering an unexhausted claim. Rather, we examine the equities of re-considering a dismissal of a claim now freed of the baggage threatening the jurisdiction of the court. The unforgiving cases cited by Texas interpret the “cause and prejudice” exception to procedural default under AEDPA and do not apply here. Significantly, AEDPA’s “cause” exception reads against fundamental principles of comity and federalism. The Supreme Court has explained that even poor lawyering cannot excuse a litigant’s failure to exhaust, because “the costs associated with an ignorant or inadvertent procedural default are no less than where the failure to raise a claim is a deliberate strategy: It deprives the state courts of the opportunity to review trial errors.”25 There is no such deprivation here. Texas has now had the opportunity to review the claim and did so on its merits. Understandably Texas cannot demonstrate why principles of comity and federalism codified in AEDPA should inform this federal court’s balancing of the equities in deciding whether to re-open its federal judgment under federal rule 60(b). With respect to Ruiz’s claims of futility, our balancing of equities here is not a question whether futility excuses non-exhaustion, it’s a question of whether futility informs our evaluation of attorney diligence. Ruiz does not argue that futility excuses non-exhaustion under AEDPA, for he has now exhausted. He is explaining only that he was not dilatory, sitting on his rights in order to weasel a last minute stay of execution; that he diligently returned to state court when that court first opened its door.

In balancing the equities, we must pay careful attention to our standard of review, which is for abuse of discretion only. The “main application” of Rule 60(b) “is to those cases in which the true merits of a *532case might never be considered.”26 Thus, although we rarely reverse a district court’s exercise of discretion to deny a Rule 60(b) motion,27 we have reversed “where denial of relief precludes examination of the full merits of the cause,” explaining that in such instances “even a slight abuse may justify reversal.”28 This lesser standard of review has been applied most liberally to motions to re-open default judgments,29 but has also been extended where a judgment on the merits was pretermitted by strict time limits in a bankruptcy court’s local rules.30 And as we have explained, no federal court has considered the merits of Ruiz’s constitutional claims. We say only that a procedural hurdle was erroneously placed in Ruiz’s path, that courts universally favor judgment on the merits, and that the underlying case here is sufficiently “significant [and] potentially meritorious” that it should not be cut off at its knees. Equity would not deny Ruiz a hearing on the merits.

The stay of execution is CONTINUED. The district court’s order denying Ruiz’s Rule 60(b) motion is REVERSED and the case is REMANDED for consideration of ineffective assistance of counsel on the merits.