These appeals were consolidated because petitioners Corpus and Vessels both claim that they are entitled to credit for “good time” served in Texas prisons pending the appeal of their criminal convictions. Vessels raises the additional issue of whether his state trial was rendered fundamentally unfair by the admission of other-crimes evidence. We deal with these issues in turn.
I GOOD-TIME CREDIT.
Approximately two years after the Texas Court of Criminal Appeals affirmed petitioners’ convictions,1 we handed down Pruett v. Texas, 470 F.2d 1182 (5th Cir.) (en banc), aff’d mem., 414 U.S. 802, 94 S.Ct. 118, 38 L.Ed.2d 39 (1973), modifying, 468 F.2d 51 (5th Cir. 1972), which held unconstitutional Texas’ practice of denying good-time credit to those held in county jails pending appeal. The en banc court expressly declared that its decision was to receive prospective application only; and were it not for a district court’s subsequent decision in Kane v. Texas, 388 F.Supp. 1188 (S.D.Tex.1975), we could dispose of petitioners’ claim simply by quoting the concluding clause of the Pruett opinion, which reads that “good time credit for the period from conviction to final conviction shall be computed and accrue as to all felony convictions which become final by affirmance of the Court of Criminal Appeals of the State of Texas after the date of this opinion [January 4, 1973].” 470 F.2d at 1184.
The holding of Kane v. Texas was that the Pruett rule should be retroactively applied to persons detained in county jails where prison officials kept conduct records adequate to enable a computation of good-time credit. The district judge based this holding on the Pruett court’s conclusion that any retroactive awards of good-time credit were administratively impossible because Texas prison officials had, in good faith, failed to maintain conduct records on inmates awaiting appellate review. Since *1380in Kane the state stipulated the adequacy of the conduct records maintained at Harris County’s correctional facilities — where Kane had been incarcerated pending the outcome of his appeal — the district court reasoned that there was nothing to bar the retroactive application of Pruett in that particular case.
Petitioners Corpus and Vessels borrow from the reasoning of Kane. Having presented evidence that some conduct records were kept in county jails before Pruett, they contend for a rule requiring a case-by-case determination of whether the state can recompute time served for each prisoner denied good-time credit because of his choice to appeal. We reject such a rule, disapprove the holding in Kane, and affirm the rulings of the courts below.
The en banc court in Pruett started with the legal conclusion that it was not required to make its ruling retroactive because the constitutional violation found in the case had nothing to do with the reliability of the guilt-determining process. 470 F.2d at 1184. A second reason, not mentioned by the court, is that the purpose of Pruett, to prohibit the state from penalizing the decision to appeal, cannot be accomplished retroactively by removing the penalty post hoc. Of course, as the en banc court noted, retroactive application “might eliminate some injustice,” 470 F.2d at 1184, and for this reason the question comes down to balancing the benefits of retroactivity against its administrative burdens. The Pruett court struck its balance, finding possible benefits outweighed by a potential flood of habeas petitions and by the absence of conduct records. While the floodwaters have presumably lessened with time and while we now know that some conduct records were kept by Texas county jails prior to the Pruett decision, we think that any rebalancing would be best left to the en banc court, this panel being bound to that court’s opinion in Pruett. Even if we were to balance the interests on our own, we would still deny petitioners’ relief. At the evidentiary hearing held below on Vessels’ petition, the state produced credible evidence that the recomputation of time served would create serious administrative difficulties because of the number of files that would have to be processed by hand. Moreover, as the lower court in Corpus’ case noted, hearings to determine pre-Pruett good-time credit would be time consuming because county prison officials did not keep conduct records in a routine and reliable fashion before the mandate in Pruett was issued.
In so holding, we reject Corpus’ claim that the district court erred in restricting his discovery of conduct records and in refusing to hold an evidentiary hearing, since we find Pruett binding regardless of the existence and adequacy of conduct records.2 Further, the extensive discovery that Corpus requested simply underscores the complexity of determining the sufficiency of conduct records.
II. OTHER-CRIMES EVIDENCE.
Vessels was convicted of rape in a state court proceeding. His victim testified at *1381trial that Vessels had come to her home posing as a gas company employee and had asked permission to check her household appliances for gas leaks. She allowed him to enter the house, and after he had carried on the charade for a few minutes, he threatened her with a knife, raped her twice, and then stabbed her several times.
The evidence to which Vessels objects was the testimony of two other women, both of whom resided in Amarillo, sixty miles from the victim’s home in Pampa. The first of these witnesses, a Mrs. Wulfman, stated that, three days before the rape, Vessels came to her home posing as a gas company employee making a routine check for gas leaks. She did not let him enter her house, however, because he lacked identification. Mrs. Wulfman also testified that her picture had appeared in a local newspaper on the morning before Vessels came to her house, just as the victim’s picture had appeared in a local newspaper a few days before her rape. The second witness, a Mrs. Hill, testified that about two weeks after the rape, Vessels broke into her home carrying a knife. During a struggle, Vessels inflicted a minor cut on her hand.
In order for us to grant habeas relief on Vessels’ evidentiary objection, the admission of the other-crimes evidence must have rendered Vessels’ trial fundamentally unfair. Hills v. Henderson, 529 F.2d 397, 401 (5th Cir.), cert. denied, 429 U.S. 850, 97 S.Ct. 139, 50 L.Ed.2d 124 (1976). We find that it did not. The testimony of Mrs. Wulfman and Mrs. Hill was admitted only to prove identity and was undoubtedly relevant to that issue, which under Texas law had been raised, albeit subtly, by defense counsel’s cross-examination of several prosecution witnesses. See Vessels v. State, 467 S.W.2d 259 (Tex.Cr.App.1971). At worst, the evidence was cumulative because of the convincing testimony given by the victim herself.3 The very strength of that testimony, however, makes it impossible for Vessels to prove that the prejudicial effect of the other-crimes evidence was a “crucial, critical, [and] highly significant factor” in his conviction, Hills v. Henderson, 529 F.2d at 401, quoting Corpus v. Beto, 469 F.2d 953, 956 (5th Cir. 1972), cert. denied, 414 U.S. 932, 94 S.Ct. 236, 38 L.Ed.2d 162 (1973), especially in light of the trial court’s limiting instruction, see Heads v. Beto, 468 F.2d 240, 241 (5th Cir. 1972), cert. denied, 410 U.S. 969, 93 S.Ct. 1454, 35 L.Ed.2d 704 (1973); and without this proof Vessels’ due process claim fails, Anderson v. Maggio, 555 F.2d 447, 451 (5th Cir. 1977).
We likewise reject Vessels’ claim that the admission of the two women’s testimony “forced” him to take the stand, which resulted in his impeachment by ten prior convictions. Whether or not a defendant should take the stand is wholly a matter of trial strategy. Vessels chose to do so and must take the consequences.
AFFIRMED.
. The Court of Criminal Appeals affirmed both convictions in 1971. For Vessels it was the second time up the appellate ladder, since three years earlier his original conviction had been reversed and his case remanded for new trial.
. For the same reason, we view as insignificant Corpus’ additional claims that county sheriffs on occasion have recommended retroactive good-time awards to inmates in county jails pending appeal and that the Texas Department of Corrections has retroactively awarded good-time credit to those held in county jails pursuant to bench warrants issued to permit temporary release from the Department for the purpose of attending judicial proceedings. Both of these practices merely suggest the existence of conduct records and thus tend to show that the administrative burdens in making retroactive awards may be less than the state contended before the en banc court in Pruett.
Corpus also argues, in a cursory fashion and without citation of authority, that the above practices violate the due process and equal protection clauses. We disagree. That one sheriff has recommended a retroactive award— which is all Corpus has shown — is of no significance in itself. As to the Department’s retroactive awards, there are several reasons for giving good-time credit to those in county jails on bench warrants and not to those awaiting appeal, including the presumably smaller number of prisoners affected and the existence of computer-filed conduct records for those already in the Department’s custody, which allows the instantaneous recomputation of time served without the necessity of contacting county officials to determine how long each prisoner was held in county jail.
. The rape occurred in mid-afternoon. The victim testified that Vessels was in her home for approximately thirty minutes and that she got a good look at his face.