On Wed, Oct 7, 2026 at 6:30 PM Alan C. Assis <[email protected]> wrote: > Hi Tomek, > I think we could solve all these Tiers issues (no pun intended) just by > keeping the HALs at github.com/NuttX because:
Tier 1 means in-tree, so apache/nuttx.git is not part of it. Tiers are here only for visualization but I like the idea it makes things simpler to explain ;-) > 1) It is still under our control; > 2) Vendors don't need to mess with HAL SGA or conversion to Apache; > 3) We avoid the supply chain issue because only PMC/Committers have write > access there. That exactly would be the difference between Tier 2 and Tier 3, as defined by SBOM, Alin can tell the details, I know this sounds like bureaucracy but all vendors will be hit by Cyber Resilience Act very soon and NuttX whether we like it or not. Zpehyr too! To be honest I was on a hands-on workshop yesterday from a very popular vendor and the guy was angry the upstream SDK versions always change in the middle of this workshops and nothing works.. plus fetching several GB if stuff from different places in different versions by 20+ people over a single internet connection makes workshop barely possible for yet another time lol. I skipped names here on purpose. This is exactly what we want to avoid in NuttX and we don't need to always follow other solutions / vendors because in the end it may turn out we were right with our beautiful coherent simplicity :-) > Anyway, if Tiago/Espressif thinks moving it to ASF is better, let's move it. > Let's fix this issue because as you said, users are waiting... > BR, > Alan Does that mean you give +1 here in this particular vote? I think we are all on the same side, and no one wants to enforce anything on anyone. If you give -1 that means there are some weak spots we need to find and solve. If all is clear we can all go +1 :-) -- CeDeROM, SQ7MHZ, http://www.tomek.cedro.info > > On Wed, Oct 7, 2026 at 3:10 PM Tomek CEDRO <[email protected]> wrote: > > > Alan, what Tiago means is that right now right here we are solving > > problem specific to Espressif. I agree with that. How other vendors > > implement their drivers is up to them. Project is under ASF, so the > > solution also should be under ASF. In a perfect situation we want NO > > HAL but in-tree drivers to stay independent. But when HAL is > > unavoidable we could have Tier 2 and Tier 3, as mentioned before: > > > > Tier 1 - in tree code drivers. > > Tier 2 - HAL that is under ASF and our control. <-- this esp-hal scenario. > > Tier 3 - HAL that is external and out of our control (not rally welcome). > > > > I can see several benefits of this situation summed up in my previous > > replies to this thread, which can be summed shortly like "we are > > moving ESP-HAL from Tier 3 to Tier 2", although we yet have no > > official tiers defined or even agreed, these are practical steps to > > fix most issues reported by Felipe. > > > > Yes, if ever any other hal is introduced (yuck!), it may be in Tier 2 > > or Tier 3. When in Tier 3 it will be in different locations anyway > > because vendors most likely prefer to keep their HALs under their own > > umbrella, you will have zero control over that code its api location > > licensing quality releases etc - this is exactly what we want to avoid > > and this is exactly why Tier 2 is being proposed right now in this > > particular vote :-) > > > > -- > > CeDeROM, SQ7MHZ, http://www.tomek.cedro.info > > > > On Wed, Oct 7, 2026 at 2:53 PM Alan C. Assis <[email protected]> wrote: > > > > > > Hi Tiago, > > > > > > Thank is not the point: "ASF has nothing to do with adopting other > > vendors' > > > HAL or not." > > > > > > If we move Espressif HAL to ASF Infra but cannot move other HALs there, > > we > > > immediately create a problem: we will spread HALs across different > > > locations. > > > > > > BR, > > > > > > Alan > > > > > > On Wed, Oct 7, 2026 at 10:51 AM Tiago Medicci Serrano < > > > [email protected]> wrote: > > > > > > > Hi all, > > > > > > > > Just to address some concerns: > > > > > > > > Tomek, a veto is valid for *code modifications.* No code is being > > > > modified here. In this case, the best fit for this vote is a > > *procedural > > > > issue *(we are moving a piece of code to ASF). That's why simple > > majority > > > > applies and I declared it as *passed* (I wasn't "pushing" for that, I > > > > based it on https://www.apache.org/foundation/voting). > > > > > > > > Alan, respectfully, all the concerns you raised are not related to this > > > > vote. ASF has nothing to do with adopting other vendors' HAL or not. > > This > > > > is a decision for the projects ASF hosts (and, of course, if they > > fulfil > > > > ASF requirements). The same for HAL becoming (or not) a standard on > > NuttX: > > > > nothing prevents native development. We voted for hosting Espressif's > > HAL > > > > under ASF (and only that). > > > > > > > > Best regards, > > > > > > > > > > > > Em qua., 7 de out. de 2026 às 10:36, Karel Kočí <[email protected]> > > > > escreveu: > > > > > > > >> Hi, > > > >> > > > >> I want to support Alan here. I would prefer the ESP HAL integration, > > when > > > >> it is > > > >> donated anyway, over HAL standardization. On this I like the Alan's > > > >> approach > > > >> more. I only feel like we might be minority. > > > >> > > > >> I am not against vendor HALs out right, I just think that they should > > be > > > >> "secondary" support and that they should not be part of the core. > > Vendors > > > >> having > > > >> option either doing the right thing and donating the code, or doing > > the > > > >> cheap > > > >> thing and just integrating with NuttX through their existing HAL. > > > >> > > > >> I am advocating for ESP integration and addition of support for > > external > > > >> chips. > > > >> > > > >> Karel > > > >> > > > >> On Wed 07 Oct 2026 10:26:30 AM , Alan C. Assis wrote: > > > >> > Hi Tomek and Tiago, > > > >> > > > > >> > Right, I can change my vote to +1 to let it pass. > > > >> > > > > >> > Although I think we are doing the wrong thing. It should have been > > done > > > >> in > > > >> > the right order. > > > >> > > > > >> > First we need to confirm with ASF if they will accept hosting any > > HAL > > > >> even > > > >> > from vendors who do not want to give a SGA to Apache. > > > >> > > > > >> > And maybe even more importantly: shouldn't we vote first on whether > > to > > > >> > accept HAL as a standard? > > > >> > > > > >> > It seems many here oppose making HAL a standard to avoid the issue > > other > > > >> > RTOSs face with poor-quality HALs. > > > >> > > > > >> > We can move the Espressif HAL to the Apache repository and over > > time, we > > > >> > can fix NuttX to replace it with native code. I think this is the > > safer > > > >> > option. > > > >> > > > > >> > BR, > > > >> > > > > >> > Alan > > > >> > > > > >> > On Wed, Oct 7, 2026 at 10:13 AM Tomek CEDRO <[email protected]> > > wrote: > > > >> > > > > >> > > Ugh, isn't that single -1 blocks the vote pass? > > > >> > > > > > >> > > Alan I think in this particular situation it is worth giving +1 > > here.. > > > >> > > would it be possible for you to reconsider and give +1 please? :-) > > > >> > > > > > >> > > -- > > > >> > > CeDeROM, SQ7MHZ, http://www.tomek.cedro.info > > > >> > > > > > >> > > On Wed, Oct 7, 2026, 15:02 Tiago Medicci Serrano < > > > >> [email protected]> > > > >> > > wrote: > > > >> > > > > > >> > > > Hi, > > > >> > > > > > > >> > > > 72 hours have already passed, so I'm closing this call for a > > vote > > > >> and its > > > >> > > > results: > > > >> > > > > > > >> > > > We had 3+ (me, Tomek, and Alin) and 1- (Alan). As a *procedural > > > >> Issue*, a > > > >> > > > simple majority is enough. So it has *passed*. > > > >> > > > > > > >> > > > We are working towards meeting ASF requirements in the coming > > days. > > > >> > > > > > > >> > > > Best regards, > > > >> > > > > > > >> > > > Em seg., 5 de out. de 2026 às 12:54, Tiago Medicci Serrano < > > > >> > > > [email protected]> escreveu: > > > >> > > > > > > >> > > > > Hi, > > > >> > > > > > > > >> > > > > As mentioned earlier, > > > >> https://github.com/espressif/esp-hal-3rdparty/ > > > >> > > is > > > >> > > > > used only by NuttX. > > > >> > > > > > > > >> > > > > Best regards, > > > >> > > > > > > > >> > > > > Em seg., 5 de out. de 2026 às 11:33, Alan C. Assis < > > > >> [email protected]> > > > >> > > > > escreveu: > > > >> > > > > > > > >> > > > >> Hi Tiago, > > > >> > > > >> > > > >> > > > >> I think we made a mistake when we said that Espressif had > > already > > > >> > > signed > > > >> > > > >> the SGA. > > > >> > > > >> > > > >> > > > >> That SGA was for software Espressif was contributing to the > > NuttX > > > >> > > > >> mainline. > > > >> > > > >> > > > >> > > > >> For the HAL, the situation is different. If Espressif signs > > an > > > >> SGA for > > > >> > > > >> their HAL it means Espressif is donating their HAL to ASF, > > the > > > >> same > > > >> > > way > > > >> > > > >> Greg donated NuttX to ASF. > > > >> > > > >> > > > >> > > > >> By doing so, Espressif will not have rights over its own HAL > > > >> (same way > > > >> > > > >> Greg > > > >> > > > >> has no rights to NuttX now). > > > >> > > > >> > > > >> > > > >> Are you sure Espressif agrees on that? If so, Espressif and > > the > > > >> Linux > > > >> > > > >> Foundation need to agree that ASF will be the owner of the > > HAL. > > > >> > > > >> > > > >> > > > >> BR, > > > >> > > > >> > > > >> > > > >> Alan > > > >> > > > >> > > > >> > > > >> On Mon, Oct 5, 2026 at 9:58 AM Tiago Medicci Serrano < > > > >> > > > >> [email protected]> wrote: > > > >> > > > >> > > > >> > > > >> > Hi, > > > >> > > > >> > > > > >> > > > >> > Just addressing the last comments: > > > >> > > > >> > > > > >> > > > >> > 1. We can't say anything about other vendors. This is how > > we > > > >> > > proposed > > > >> > > > to > > > >> > > > >> > host Espressif's HAL: under the Apache umbrella because we > > > >> already > > > >> > > > have > > > >> > > > >> a > > > >> > > > >> > signed CCLA. *Keep in mind that this call for a vote is > > about > > > >> > > > >> Espressif's > > > >> > > > >> > HAL only*. > > > >> > > > >> > 2. It's being contributed to a separate repository because: > > > >> > > > >> > 2.1 It eases new contributions and new syncs being > > contributed > > > >> by > > > >> > > > >> > Espressif. > > > >> > > > >> > 2.2 The HAL has its own coding style and contribution > > guide. > > > >> > > > >> > 2.3 It makes it easier to implement native implementations > > and > > > >> > > isolate > > > >> > > > >> them > > > >> > > > >> > from the HAL. > > > >> > > > >> > > > > >> > > > >> > Please, keep voting on the proposed subject. Nothing else > > is > > > >> being > > > >> > > > >> > discussed here (we have other threads to discuss it). > > > >> > > > >> > > > > >> > > > >> > Best regards, > > > >> > > > >> > > > > >> > > > >> > Em sáb., 3 de out. de 2026 às 09:11, Alan C. Assis < > > > >> > > [email protected] > > > >> > > > > > > > >> > > > >> > escreveu: > > > >> > > > >> > > > > >> > > > >> > > Hi Tomek, > > > >> > > > >> > > > > > >> > > > >> > > These are the issues I'm seeing here: > > > >> > > > >> > > > > > >> > > > >> > > 1) If Espressif donates its HAL to ASF it means they > > will no > > > >> > > longer > > > >> > > > >> own > > > >> > > > >> > the > > > >> > > > >> > > HAL (you cannot have your cake and eat it too). Is this > > > >> acceptable > > > >> > > > to > > > >> > > > >> > them? > > > >> > > > >> > > > > > >> > > > >> > > 2) Even if they accept this, other vendors probably will > > not > > > >> > > accept > > > >> > > > >> > > donating their HALs. In this case we cannot keep it under > > > >> ASF. > > > >> > > > >> > > If so, we would have to place it elsewhere, which would > > cause > > > >> > > > >> confusion: > > > >> > > > >> > > why are some HALs under ASF and others in a different > > > >> location? > > > >> > > > >> > > > > > >> > > > >> > > BR, > > > >> > > > >> > > > > > >> > > > >> > > Alan > > > >> > > > >> > > > > > >> > > > >> > > On Fri, Oct 2, 2026 at 10:27 PM Tomek CEDRO < > > > >> [email protected]> > > > >> > > > wrote: > > > >> > > > >> > > > > > >> > > > >> > > > Hey there Alan, > > > >> > > > >> > > > > > > >> > > > >> > > > Most of the raised HAL related issues should be solved > > > >> that way > > > >> > > as > > > >> > > > >> > > > proposed by Tiago and Espressif in this vote: > > > >> > > > >> > > > 1. Code ownership and supply chain security (mentioned > > by > > > >> Alin). > > > >> > > > >> > > > 2. Community involvement in faster PR processing > > > >> (mentioned by > > > >> > > > >> Felipe). > > > >> > > > >> > > > 3. We should have more control and impact on the HAL > > > >> quality and > > > >> > > > >> > > > availability (mentioned by many). > > > >> > > > >> > > > 4. HAL is under the NuttX umbrella just as proposed > > "Zephyr > > > >> > > > Umbrella > > > >> > > > >> > > > Way" (mentioned by Renesas). > > > >> > > > >> > > > > > > >> > > > >> > > > From I understand Espressif offers to "donate" HAL to > > ASF > > > >> to be > > > >> > > > >> > > > apache/nuttx-hal-espressif and this seems the best > > possible > > > >> > > > win-win > > > >> > > > >> > > > scenario. They don't have to do this but they want - > > big > > > >> thank > > > >> > > > you! > > > >> > > > >> > > > > > > >> > > > >> > > > What obstacles do you see here? The vote here is about > > > >> "Hosting > > > >> > > > >> > > > Espressif HAL under Apache umbrella" not all HALs. We > > have > > > >> > > perfect > > > >> > > > >> > > > road paving here and good example for others. If we > > have > > > >> problem > > > >> > > > >> with > > > >> > > > >> > > > other HAL then we vote other thing. But why block good > > > >> solution > > > >> > > > >> > > > because of other possible bad solutions? > > > >> > > > >> > > > > > > >> > > > >> > > > We may one day consider Tiers as proposed by Michal and > > > >> > > mentioned > > > >> > > > by > > > >> > > > >> > > > Nathan (with FreeBSD): > > > >> > > > >> > > > 1. Tier 1 - in tree code drivers. > > > >> > > > >> > > > 2. Tier 2 - HAL that is under ASF and our control. <-- > > we > > > >> are > > > >> > > > voting > > > >> > > > >> > > > this scenario here. > > > >> > > > >> > > > 3. Tier 3 - HAL that is external and out of our control > > > >> (not > > > >> > > rally > > > >> > > > >> > > > welcome). > > > >> > > > >> > > > > > > >> > > > >> > > > Could you please reconsider your vote for this > > particular > > > >> case / > > > >> > > > >> > > > scenario that is being voted? It seems that Espressif > > > >> listened > > > >> > > > >> > > > carefully and provides best possible scenario :-) > > > >> > > > >> > > > > > > >> > > > >> > > > Thanks :-) > > > >> > > > >> > > > Tomek > > > >> > > > >> > > > > > > >> > > > >> > > > -- > > > >> > > > >> > > > CeDeROM, SQ7MHZ, http://www.tomek.cedro.info > > > >> > > > >> > > > On Fri, Oct 2, 2026 at 3:10 PM Alan C. Assis < > > > >> [email protected] > > > >> > > > > > > >> > > > >> > wrote: > > > >> > > > >> > > > > > > > >> > > > >> > > > > -1 > > > >> > > > >> > > > > > > > >> > > > >> > > > > As I explained before we cannot keep all HALs under > > ASF. > > > >> > > > >> > > > > > > > >> > > > >> > > > > BR, > > > >> > > > >> > > > > > > > >> > > > >> > > > > Alan > > > >> > > > >> > > > > > > > >> > > > >> > > > > On Fri, Oct 2, 2026 at 10:31 AM Tiago Medicci > > Serrano < > > > >> > > > >> > > > > [email protected]> wrote: > > > >> > > > >> > > > > > > > >> > > > >> > > > > > Hi All, > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > After several discussions about HAL usage and > > community > > > >> > > > >> governance > > > >> > > > >> > > > > > [1][2][3], we (Espressif) propose hosting the > > current > > > >> > > content > > > >> > > > >> of ` > > > >> > > > >> > > > > > github.com/espressif/esp-hal-3rdparty` > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > > > > > <http://github.com/espressif/esp-hal-3rdparty> > > under > > > >> an > > > >> > > > >> ASF-hosted > > > >> > > > >> > > > > > repository: ` > > > >> > > > >> > > > > > github.com/apache/nuttx-hal-espressif` > > <http://github.com/apache/nuttx-hal-espressif> > > > >> <http://github.com/apache/nuttx-hal-espressif> > > > >> > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > > > > > <http://github.com/apache/nuttx-hal-espressif>. > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > *What does it mean:* > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > - `esp-hal-3rdparty` content is already Apache > > 2.0 > > > >> > > > licensed. > > > >> > > > >> > > > > > - Espressif's CCLA allows us to contribute to a > > > >> newly > > > >> > > > created > > > >> > > > >> > > > > > repository: we would create a PR on ` > > > >> > > > >> > > > > > github.com/apache/nuttx-hal-espressif` > > <http://github.com/apache/nuttx-hal-espressif> > > > >> <http://github.com/apache/nuttx-hal-espressif> > > > >> > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > > > <http://github.com/apache/nuttx-hal-espressif> > > > >> > > > >> > > > > > <http://github.com/apache/nuttx-hal-espressif> > > with > > > >> > > > >> > > `esp-hal-3rdparty` > > > >> > > > >> > > > > > content. > > > >> > > > >> > > > > > - On the other hand, the org account on ` > > > >> > > > >> github.com/nuttx` <http://github.com/nuttx> < > > http://github.com/nuttx> < > > > >> http://github.com/nuttx> <http://github.com/nuttx> > > > >> > > <http://github.com/nuttx> > > > >> > > > >> > <http://github.com/nuttx> > > > >> > > > >> > > <http://github.com/nuttx> > > > >> > > > >> > > > <http://github.com/nuttx> > > > >> > > > >> > > > > > <http://github.com/nuttx> is not > > > >> > > > >> > > > > > legally attached to Apache Software > > Foundation > > > >> and, > > > >> > > > >> > > > > > consequently, to NuttX. > > > >> > > > >> > > > > > We can't directly contribute to it. > > > >> > > > >> > > > > > - NuttX's committers will have (just like > > `nuttx` > > > >> and > > > >> > > > >> > > `nuttx-apps`) > > > >> > > > >> > > > > > write permissions to merge PRs on the newly > > created > > > >> > > > >> repository; > > > >> > > > >> > > > > > - Espressif will support the community with HAL > > > >> usage, > > > >> > > > >> > including a > > > >> > > > >> > > > > > comprehensive guide on how to update it so that > > any > > > >> > > > community > > > >> > > > >> > > > member can > > > >> > > > >> > > > > > send patches for HAL. > > > >> > > > >> > > > > > - The discussion about HAL governance comes and > > > >> goes, > > > >> > > > >> creating > > > >> > > > >> > > > > > instability in the NuttX community among > > members, > > > >> > > > developers, > > > >> > > > >> > and > > > >> > > > >> > > > > > vendors. > > > >> > > > >> > > > > > This is no longer acceptable and needs > > settlement. > > > >> That > > > >> > > > being > > > >> > > > >> > > said: > > > >> > > > >> > > > > > - `github.com/espressif/esp-hal-3rdparty` > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > > > <http://github.com/espressif/esp-hal-3rdparty> > > > >> > > > >> > > > > > <http://github.com/espressif/esp-hal-3rdparty> > > will no > > > >> > > longer > > > >> > > > >> > > receive > > > >> > > > >> > > > > > direct contributions for now. > > > >> > > > >> > > > > > - The newly created > > `apache/nuttx-hal-espressif` > > > >> - if > > > >> > > > the > > > >> > > > >> > > > community > > > >> > > > >> > > > > > agrees on that - will enable it to evolve and > > > >> receive > > > >> > > > new > > > >> > > > >> > > > > > contributions > > > >> > > > >> > > > > > with more authority from the community. > > > >> > > > >> > > > > > - On the other hand, native development is > > > >> always an > > > >> > > > >> option > > > >> > > > >> > if > > > >> > > > >> > > > the > > > >> > > > >> > > > > > community decides to keep supporting the > > devices > > > >> on > > > >> > > its > > > >> > > > >> own. > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > *What it doesn't mean:* > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > - It doesn't mean we are pushing NuttX to accept > > > >> vendor > > > >> > > > HALs. > > > >> > > > >> > > > > > `esp-hal-3rdparty` is a reality, and we want to > > > >> empower > > > >> > > the > > > >> > > > >> > > > community to > > > >> > > > >> > > > > > enable any NuttX member to directly work with > > it. > > > >> > > > >> > > > > > - The discussion of whether NuttX should or > > > >> should not > > > >> > > > >> > > officially > > > >> > > > >> > > > > > approve vendor HALs is not related to this > > call > > > >> for a > > > >> > > > >> vote. > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > This seems to be a *Procedural Issue* [4], and a > > simple > > > >> > > > majority > > > >> > > > >> > > > should be > > > >> > > > >> > > > > > applied. The community is encouraged to vote. PMC > > > >> members > > > >> > > have > > > >> > > > >> > > binding > > > >> > > > >> > > > > > votes. > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > In that sense, my vote (Tiago) is *+1 *(binding) > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > References: > > > >> > > > >> > > > > > [1] > > > >> > > > >> > > > > >> https://lists.apache.org/thread/gf4c3z4ds94lon91op4c6zmclqnwtvcf > > > >> > > > >> > > > > > [2] > > > >> > > > >> > > > > >> https://lists.apache.org/thread/nvnrf30jcdsxcrydmo2s95tdtlj8zvx9 > > > >> > > > >> > > > > > [3] > > > >> > > > >> > > > > >> https://lists.apache.org/thread/bg74qn14rs9qwson5lzt0xjb4fvbqh6j > > > >> > > > >> > > > > > [4] https://www.apache.org/foundation/voting > > > >> > > > >> > > > > > > > > >> > > > >> > > > > > > >> > > > >> > > > > > >> > > > >> > > > > >> > > > >> > > > >> > > > > > > > >> > > > > > > >> > > > > > >> > > > > > >
