Analayo indeed reviewed Yinshun work. In A Comparative Study of the Majjhima-nikaya vol. 2 page 179 footnote 69, he wrote a long comment of Yinshun findings:
Yìnshùn 1971/1983: 788 considers the three angas of discourse, verses, and exposition,
修多羅, 祇夜, 記說, to be the earliest organizing principle in the evolution of the early Buddhist canon. Choong 2010: 56-57 explains that according to this hypothesis these three angas underlie the structure of the Samyukta-agama, considered to be “the foundation of all four agamas (nikayas) in the formation of early Buddhist texts”, which then “subsequently expanded and yielded the other agamas in the sequence Madhyama-agama, Dirgha-agama, Ekottarika-agama. Therefore, the … SA [Samyukta-agama] is the foundation of all four agamas”, in contrast to “the secondary nature of [the] Madhyamagama/Majjhima-nikaya, Dirghagama/Digha-nikaya, and Ekottarikagama/Anguttara-nikaya”. This hypothesis is based on the description of the Agamas given in the Vastusamgrahani of the Yogacarabhumi.Although the indications made in the Vastusamgrahani have indubitably been of great importance for reconstructing the order of the Samyukta-agama (T 99), which in the Taisho edition is in disarray (cf. also Bucknell 2006: 685), from a methodological viewpoint it would not be possible to use the reconstructed Samyukta-agama in turn to prove that the indications given in the Vastusamgrahani are correct, since this would become a circular argument. Besides, it is also not clear if the description of the Agamas in the Vastusamgrahani has to be read as positing the Samyukta-agama as a kind of ‘Urkanon’ in its own right. It might just intend to explain why the Samyukta-agama is given pride of place in its listings of the four Agamas. In the corresponding passage in the (Mula-)Sarvastivada Vinaya, T 1451 at T XXIV 407b27, cf. also Lévi 1916: 36, the point at stake does in fact not seem to be a temporal priority of the formation of the Samyukta-agama, but only a temporal priority of it being recited by Ananda at the so-called first council, before he recited the other Agamas.
Although the idea that the early Buddhist oral transmission began by assembling discourses according to topics in a manner similar to what is not found in the Samyukta-agama and the Samyutta-nikaya is certainly appealing, there seems to be no reason why a numerical organisation could not have been in use as well, similar to what underlies the Ekottarika-agama and the Anguttara-nikaya. The introduction to the Ekottarika-agama in fact emphasizes this numerical arrangement, T II 550b24, and hence gives pride of place among the four collections to the Ekottarika-agama, with the Samyukta-agama relegated to the last position in its list, cf. T II 549c29 (other listings that place the Ekottarika-agama in first position and the Samyukta-agama last can be found, e.g., in the (presumably) Haimavata Vinaya, T 1463 at T XXIV 820a23, the *Mahaprajñaparamita-(upadesa-)sastra, T 1509 at T XXV 69c5, and the 入大乘論 , T 1634 at T XXXII 36c15; for still other listings cf. below p. 864 note 46). Perhaps more than one organizing principle was in use from the outset, since otherwise it would be difficult to explain what happened to those discourses that do not neatly fit into the topic-wise arrangement now found in the Samyukta-agama/Samyutta-nikaya.
It seems improbable that these were just left to float around without being in any way organized at a time when other discourses were formally assembled according to topic. Alternatively, all such discourses must be considered to be of later origin, resulting in a presumption about the nature of earliest Buddhism that would be difficult to substantiate. The Samyukta-agama (T 99) itself contains a number of passages and tales that in the Theravada tradition are found only in commentarial literature. This makes it improbable that the Samyukta-agama, at least in the form we have it now, should be assigned to a textual strata that is distinctively earlier than the other discourse collections (cf. also Choong 2010: 63, who notes that the discourse material found in this collection does not “all actually belong to the teachings of Early Buddhism”).
Regarding the angas, the description of the Agamas in the passage under discussion from the Vastusamgrahani does not explicitly refer to the set of three angas. The only tri-partite analysis found in the present passage distinguishes the Samyukta-agama discourses from the viewpoint of speaker, topic, and audience, T 1579 at T XXX 772c17: 一是能說, 二是所說, 三是所為說 and D (4039) sems tsam, zi 128a1 or Q (5540) ’i 143b6: su ston pa dang, ci ston pa dang, gang la ston pa dang (cf. also Bucknell 2007: 19 and 32 note 78). This division does not naturally evoke the three angas of sutta, geyya, and veyyakarana (on the significance of the last of these three cf. also Analayo 2009l). In fact, had this been the original intention, the three angas could have been directly mentioned.
Another argument in support of the significance of the threefold listing is presented by Sujato 2005: 62, who notes that in a reference to the twelve angas in a Mahaparinirvana-sutra fragment, cf. SN 362 folio 173R6-7 in Waldschmidt 1950: 62, “the twelve angas are listed with the first three members declined as individual words, while the remaining members occur grouped together as one long compound”, which he takes to point to a special emphasis on the first three. Although in the case of another Mahaparinirva0a-sutra version, fragment 1024R5 in Waldschmidt 1968: 5, the remainder of the listing does not form a compound: sutram geyam vyakaranam gath-oddana nidanavadana itivr(ttkajataka)v[ai]pulyadbhutadharmopa(desas), Sujato notes that in the Sravakabhumi listings of the angas also follow the above pattern, cf., e.g., Shukla 1973: 100,18 or SSG 1998: 154,3. Yet, this form of presentation need not be interpreted as reflecting some underlying awareness of an ancient use of the three angas at the time of the formation of the canonical collections, otherwise forgotten.
It could just be the result of a standard procedure in Buddhist texts, where the first three members of listing are always mentioned, even when the rest is abbreviated (cf. also note 70 below). This pattern can in fact be observed in the same Sravakabhumi, cf., e.g., Shukla 1973: 101,20 or SSG 1998: 154,25: sutram geyan vyakaranam iti vistarena purvavat, a formulation that could easily have influenced the pattern noticed by Sujato. In sum, it seems that the evidence brought forward so far in support of a special significance of the three angas is not conclusive. Hence, the present passage in MA 191 (and presumably also in MN 122) might, after all, be just the result of a transmission error.